Voices in AI – Episode 71: A Conversation with Paul Daugherty

[voices_in_ai_byline]

About this Episode

Episode 71 of Voices in AI features host Byron Reese and Paul Daugherty discuss transfer learning, consciousness and Paul’s book “Human + Machine: Reimagining Work in the Age of AI.” Paul Daugherty holds a degree in computer engineering from the University of Michigan, and is currently the Chief Technology and Innovation Officer at Accenture.
Visit www.VoicesinAI.com to listen to this one-hour podcast or read the full transcript.

Transcript Excerpt

Byron Reese: This is Voices in AI brought to you by GigaOm. Today my guest is Paul Daugherty. He is the Chief Technology and Innovation Officer at Accenture. He holds a computer engineering degree from the University of Michigan. Welcome to the show Paul.

Paul Daugherty: It’s great to be here, Byron.

Looking at your dates on LinkedIn, it looks like you went to work for Accenture right out of college and that was a quarter of a century or more ago. Having seen the company grow… What has that journey been like?

Thanks for dating me. Yeah it’s actually been 32 years, so I guess I’m going on a third of a century, joined Accenture back in 1986, and the company’s evolved in many ways since then. It’s been an amazing journey because the world has changed so much since then and a lot of what’s fueled the change in the world around us has been what’s happened with technology. I think [in] 1986 the PC was brand new, and we went from that to networking and client server and the Internet, cloud computing mobility, internet of things, artificial intelligence and the things we’re working on today. So it’s been a really amazing journey fueled by the way the world’s changed, enabled by all this amazing technology.

So let’s talk about that, specifically artificial intelligence. I always like to get our bearings by asking you to define either artificial intelligence or if you’re really feeling bold, define intelligence.

I’ll start with artificial intelligence which we define as technology that can sense, think, act and learn, is the way we describe it. And [it’s] systems that can then do that, so sense: like vision in a self-driving car, think: making decisions on what the car does next, acts: in terms of they actually steer the car and then learn: to continuously improve behavior. So that’s the working definition that we use for artificial intelligence, and I describe it more simply to people sometimes, as fundamentally technology that has more human-like capability to approximate the things that we’re used to assuming and thinking that only humans can do: speech, vision, predictive capability and some things like that.

So that’s the way I define artificial intelligence. Intelligence I would define differently. Intelligence I would just define more broadly. I’m not an expert in neuroscience or cognitive science or anything, but I define intelligence generally as the ability to both reason and comprehend and then extrapolate and generalize across many different domains of knowledge. And that’s what differentiates human intelligence from artificial intelligence, which is something we can get a lot more into. Because I think the fact that we call this body of work that we’re doing artificial intelligence, both the word artificial and the word intelligence I think lead to misleading perceptions on what we’re really doing.

So, expand that a little bit. You said that’s the way you think human intelligence is different than artificial, — put a little flesh on those bones, in exactly what way do you think it is?

Well, you know the techniques we’re really using today for artificial intelligence, they’re generally from the branch of AI around machine learning, so machine learning, deep learning, neural nets etc. And it’s a technology that’s very good at using patterns and recognizing patterns in data to learn from observed behavior, so to speak. Not necessarily intelligence in a broad sense, it’s ability to learn from specific inputs. And you can think about that almost as idiot savant-like capability.

So yes, I can use that to develop Alpha Go to beat the world’s Go master, but then that same program wouldn’t know how to generalize and play me in tic-tac-toe. And that ability, the intelligence ability to generalize, extrapolate, rather than interpolate, is what human intelligence is differentiated by, and the thing that would bridge that, would be artificial general intelligence, which we can get into a little bit, but we’re not at that point of having artificial general intelligence, we’re at a point of artificial intelligence, where it could mimic very specific, very specialised, very narrow human capabilities, but it’s not yet anywhere close to human-level intelligence.

Listen to this one-hour episode or read the full transcript at www.VoicesinAI.com
[voices_in_ai_link_back]
 
Byron explores issues around artificial intelligence and conscious computers in his new book The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity.

Voices in AI – Episode 62: A Conversation with Atif Kureishy

[voices_in_ai_byline]

About this Episode

Episode 62 of Voices in AI features host Byron Reese and Atif Kureishy discussing AI, deep learning, and the practical examples and implications in the business market and beyond. Atif Kureishy is the Global VP of Emerging Practices at Think Big, a Teradata company. He also has a B.S. in physics and math from the University of Maryland as well as an MS in distributive computing from Johns Hopkins University.
Visit www.VoicesinAI.com to listen to this one-hour podcast or read the full transcript.

Transcript Excerpt

Byron Reese: This is Voices in AI, brought to you by GigaOm, I’m Byron Reese. Today my guest is Atif Kureishy. He is the Global VP of Emerging Practices, which is AI and deep learning at Think Big, a Teradata company. He holds a BS in Physics and Math from the University of Maryland, Baltimore County, and an MS in distributive computing from the Johns Hopkins University. Welcome to the show Atif.
Atif Kureishy: Welcome, thank you, appreciate it.
So I always like to start off by just asking you to define artificial intelligence.
Yeah, definitely an important definition, one that unfortunately is overused and stretched in many different ways. Here at Think Big we actually have a very specific definition within the enterprise. But before I give that, for me in particular, when I think of intelligence, that conjures up the ability to understand, the ability to reason, the ability to learn, and we usually equate that to biological systems, or living entities, and now with the rise of probably more appropriate machine intelligence, we’re applying the term ‘artificial’ to it, and the rationale is probably because machines aren’t living and they’re not biological systems.
So with that, the way we’ve defined AI in particular is: leveraging machine and deep learning to drive towards a specific business outcome. And it’s about giving leverage for human workers, to enable higher degrees of assistance and higher degrees of automation. And when we define AI in that way, we actually give it three characteristics. Those three characteristics are: the ability to sense and learn, and so that’s being able to understand massive amounts to data and demonstrate continuous learning, and detecting patterns and signals within the noise, if you will. And the second is being able to reason and infer, and that is driving intuition and inference with increasing accuracy again to maximize a business outcome or a business decision. And then ultimately it’s about deciding and acting, so actioning or automating a decision based on everything that’s understood, to drive towards more informed activities that are based on corporate intelligence. So that’s kind of how we view AI in particular.
Well I applaud you for having given it so much thought, and there’s a lot there to unpack. You talked about intelligence being about understanding and reasoning and learning, and that was even in your three areas. Do you believe machines can reason?
You know, over time, we’re going to start to apply algorithms and specific models to the concept of reasoning, and so the ability to understand, the ability to learn, are things that we’re going to express in mathematical terms no doubt. Does it give it human lifelike characteristics? That’s still something to be determined.
Well I don’t mean to be difficult with the definition because, as you point out, most people aren’t particularly rigorous when it comes to it. But if it’s to drive an outcome, take a cat food dish that refills itself when it’s low, it can sense, it can reason that it should put more food in, and then it can act and release a mechanism that refills the food dish, is that AI, in your understanding, and if not why isn’t that AI?
Yeah, I mean I think in some sense it checks a lot of the boxes, but the reality is, being able to adapt and understand what’s occurring, for instance if that cat is coming out during certain times of the day ensuring that meals are prepared in the right way and that they don’t sit out and become stale or become spoiled in any way, and that is signs of a more intelligent type of capability that is learning the behaviors and anticipating how best to respond given a specific outcome it’s driving towards.
Got you. So now, to take that definition, your company is Think Big. What do you think big about? What is Think Big and what do you do?
So looking back in history a little bit, Think Big was actually an acquisition that Teradata had done several years ago, in the big data space, and particularly around open source and consulting. And over time, Teradata had made several acquisitions and now we’ve unified all of those various acquisitions into a unified group, called Think Big Analytics. And so what we’re particularly focused on is how do we drive business outcomes using advanced analytics and data science. And we do that through a blend of approaches and techniques and technology frankly.
Listen to this one-hour episode or read the full transcript at www.VoicesinAI.com
[voices_in_ai_link_back]
 
Byron explores issues around artificial intelligence and conscious computers in his new book The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity.

Voices in AI – Episode 50: A Conversation with Steve Pratt

[voices_in_ai_byline]
In this episode, Byron and Steve discuss the present and future impact of AI on businesses.
[podcast_player name=”Episode 50: A Conversation with Steve Pratt” artist=”Byron Reese” album=”Voices in AI” url=”https://voicesinai.s3.amazonaws.com/2018-06-14-(00-56-12)-stephen-pratt.mp3″ cover_art_url=”https://voicesinai.com/wp-content/uploads/2018/06/voices-headshot-card-3.jpg”]
[voices_in_ai_byline]
Byron Reese: This is Voices in AI, brought to you by GigaOm, and I’m Byron Reese. Today, our guest is Steve Pratt. He is the Chief Executive Officer over at Noodle AI, the enterprise artificial intelligence company. Prior to Noodle, he was responsible for all Watson implementations worldwide, for IBM Global Business Services. He was also the founder and CEO of Infosys Consulting, a Senior Partner at Deloitte Consulting, and a Technology and Strategy Consultant at Booz Allen Hamilton. Consulting Magazine has twice selected him as one of the top 25 consultants in the world. He has a Bachelor’s and a Master’s in Electrical Engineering from Northwestern University and George Washington University. Welcome to the show, Steve.
Steve Pratt: Thank you. Great to be here, Byron.
Let’s start with the basics. What is artificial intelligence, and why is it artificial?
Artificial intelligence is basically any form of learning algorithm; is the way we think of things. We actually think there’s a raging religious debate [about] the differences between artificial intelligence and machine learning, and data science, and cognitive computing, and all of that. But we like to get down to basics, and basically say that they are algorithms that learn from data, and improve over time, and are probabilistic in nature. Basically, it’s anything that learns from data, and improves over time.
So, kind of by definition, the way that you’re thinking of it is it models the future, solely based on the past. Correct?
Yes. Generally, it models the future and sometimes makes recommendations, or it will sometimes just explain things more clearly. It typically uses four categories of data. There is both internal data and external data, and both structured and unstructured data. So, you can think of it kind of as a quadrant. We think the best AI algorithms incorporate all four datasets, because especially in the enterprise, where we’re focused, most of the business value is in the structured data. But usually unstructured data can add a lot of predictive capabilities, and a lot of signal, to come up with better predictions and recommendations.
How about the unstructured stuff? Talk about that for a minute. How close do you think we are? When do you think we’ll have real, true unstructured learning, that you can kind of just point at something and say, “I’m going to Barbados. You figure it all out, computer.”
I think we have versions of that right now. I am an anti-fan of things like chatbots. I think that chatbots are very, very difficult to do, technically. They don’t work very well. They’re generally very expensive to build. Humans just love to mess around with chatbots. I would say in the scoring of business value and something that’s affordable, and is easy to do, that chatbots is in the worst quadrant there.
I think there is a vast array of other things that actually add business value to companies, but if you want to build an intelligent agent using natural language processing, you can do some very basic things. But I wouldn’t start there.
Let me try my question slightly differently, then. Right now, the way we use machine learning is we say, “We have this problem that we want to solve. How do you do X?” And we have this data that we believe we can tease the answer out of. We ask the machine to analyze the data, and figure out how to do that. It seems the inherent limit of that, though, it’s kind of all sequential in nature. There’s no element of transferred learning in that, where I grow exponentially what I’m able to do. I just can do: “Yes. Another thing. Yes. Another. Yes. Another.” So, do you think this strict definition of machine learning, as you’re thinking of AI that way, is that a path to a general intelligence? Or is general intelligence like “No, that’s something way different than what we’re trying to do. We’re just trying to drive a car, without hitting somebody?”
General intelligence, I think, is way off in the future. I think we’re going to have to come up with some tremendous breakthroughs to get there. I think you can duct-tape together a lot of narrow intelligence, and sort of approximate general intelligence, but there are some fundamental skills that computers just can’t do right now.For instance, if I give a human the question, “Will the guinea pig population in Peru be relevant to predicting demand for tires in the U.S?” A human would say, “No, that’s silly. Of course not.” A computer would not know that. A computer would actually have to go through all of the calculations, and we don’t have an answer to that question, yet. So, I think generalized intelligence is a way off, but I think there are some tremendously exciting things that are happening right now, that are making the world a better place, in narrow intelligence.
Absolutely. I do want to spend the bulk of our time in there, in that world. But just to explore what you were saying, because there’s a lot of stuff to mine, in what you just said. That example you gave about the guinea pigs is sort of a common-sense problem, right? In how it’s referred. “Am I heavier than the statue of liberty?” How do you think humans are so good at that stuff? How is it that if I said, “Hey, what would an Oscar statue look like, smeared with peanut butter?” You can conjure that up, even though you’ve never even thought of that before, or seen it covered, or seen anything covered with peanut butter. Why are we so good at that kind of stuff, and machines seem amazingly ill-equipped at it?
I think humans have constant access to an incredibly diverse array of datasets. Through time, they have figured out patterns from all of those diverse datasets. So, we are constantly absorbing new datasets. In machines, it’s a very deliberate and narrow process right now. When you’re growing up, you’re just seeing all kinds of things. And as we go through our life, we develop these – you could think of them as regressions and classifications in our brains, for those vast arrays of datasets.
As of right now, machine learning and AI are given very specific datasets, crunch the data, and then make a conclusion. So, it’s somewhere in there. We’re not exactly sure, yet.
All right, last question on general intelligence, and we’ll come back to the here and now. When I ask people about it, the range of answers I get is 5 to 500 years. I won’t pin you down to a time, but it sounds like you’re “Yeah, it’s way off.” Yet, people who say that often usually say, “We don’t know how to do it, and it’s going to be a long time before we get it.”
But there’s always the implicit confidence that we can do it, that it is a possible thing. We don’t know how to do it. We don’t know how we’re intelligent. We don’t know the mechanism by which we are conscious, or the mechanism by which we have a mind, or how the brain fundamentally functions, and all of that. But we have a basic belief that it’s all mechanistic, so we’re going to eventually be able to build it. Do you believe that, or is it possible that a general intelligence is impossible?
No. I don’t think it’s impossible, but we just don’t know how to do it, yet. I think transfer learning, there’s a clue in there, somewhere. I think you’re going to need a lot more memory, and a lot more processing power, to have a lot more datasets in general intelligence. But I think it’s way off. I think there will be stage gates, and there will be clues of when it’s starting to happen. That’s when you can take an algorithm that’s trained for one thing, and have it – if you can take Alpha Go, and then the next day, it’s pretty good at Chess. And the next day, it’s really good at Parcheesi, and the next day, it’s really good at solving mazes, then we’re on the track. But that’s a long way off.
Let’s talk about this narrow AI world. Let’s specifically talk about the enterprise. Somebody listening today is at, let’s say a company of 200 people, and they do something. They make something, they ship it, they have an accounting department, and all of that. Should they be thinking about artificial intelligence now? And if so, how? How should they think about applying it to their business?
A company that small, it’s actually really tough, because artificial intelligence really comes into play when it’s beyond the complexity that a human can fit in their mind.
Okay. Let’s up it to 20,000 people.
20,000? Okay, perfect. 20,000 people – there are many, many places in the organization where they absolutely should be using learning algorithms to improve their decision-making. Specifically, we have 5 applications that focus on the supply side of the company; that’s in: materials, production, distribution, logistics and inventory.
And then, on the supply side, we have 5 areas also: customer, product, price, promotion and sales force. All of those things are incredibly complex, and they are highly interactive. Within each application area, we basically have applications that almost treat it like a game, although it’s much more complicated than a game, even though games like Go are very complex.
Each of our applications does, really, 4 things: it senses, it proposes, it predicts, and then it scores. So, basically it senses the current environment, it proposes a set of actions that you could take, it predicts the outcome of each of those actions – like the moves on a Chessboard – and then it scores it. It says, “Did it improve?” There are two levels of that, two levels of sophistication. One is “Did it improve locally? Did it improve your production environment, or your logistics environment, or your materials environment?” And then, there is one that is more complex, which says “If you look at that across the enterprise, did it improve across the enterprise?” These are very, very complex mathematical challenges. The difference is dramatic, from the way decisions are made today, which is basically people getting in meetings with imperfect data on spreadsheets and PowerPoint slides, and having arguments.
So, pick a department, and just walk me through a hypothetical or real use case where you have seen the technology applied, and have measurable results.
Sure. I can take the work we’re doing at XOJET, which is the largest private aviation company in the U.S. If you want to charter a jet, XOJET is the leading company to do that. The way they were doing pricing before we got there was basically old, static rules that they had developed several years earlier. That’s how they were doing pricing. What we did is we worked with them to take into account where all of their jets currently were, where all of their competitors’ jets are, what the demand was going to be, based on a lot of internal and external data; like what events were happening in what locations, what was the weather forecast, what [were] the economic conditions, what were historic prices and results? And then, basically came up with all of the different pricing options they could come up with, and then basically made a recommendation on what the price should be. As soon as they put in our application, which was in Q4 of 2016, the EBITDA of the company, which is basically the net margin – not quite, but – went up 5%, in the company.
The next thing we did for them was to develop an application that looked at the balance in their fleet, which is: “Do you have the right jets in the right place, at the right time?” This takes into account having to look at the next day. Where is the demand going to be the next day? So, you make sure you don’t have too many jets in low demand locations, or not enough jets in high demand locations. We actually adjusted the prices, to create an economic incentive to drive the jets to the right place at the right time.
We also, again, looked at competitive position, which is through Federal Aviation Administration data. You can track the tail numbers of all of their jets, and all of the competitor jets, so you could calculate competitive position. Then, based on that algorithm, the length of haul, which is the amount of hours flown per jet, went up 11%.
This was really dramatic, and dramatically reduced the number of “deadheads” they were flying, which is the amount of empty jets they were flying to reposition their jets. I think that’s a great success story. There’s tremendous leadership at that company, very innovative, and I think that’s really transformed their business.
That’s kind of a classic load-balancing problem, right? I’ve got all of these things, and I want to kind of distribute it, and make sure I have plenty of what I need, where. That sounds like a pretty general problem. You could apply it to package delivery or taxicab distribution, or any number of other things. How generalizable is any given solution, like from that, to other industries?
That’s a great question. There are a lot of components in that, that are generalizable. In fact, we’ve done that. We have componentized the code and the thinking, and can rapidly reproduce applications for another client, based on that. There’s a lot of stuff that’s very specific to the client, and of course, the end application is trained on the client’s data. So, it’s not applicable to anybody else. The models are specifically trained on the client data. We’re doing other projects in airline pricing, but the end result is very different, because the circumstances are different.
But you hit on a key question, which is “Are things generalizable?” One of the other approaches we’re taking is around transferred learning, especially when you’re using deep learning technologies. You can think of it as the top layers of a neural net can be trained on sort of general pricing techniques, and just the deeper layers are trained on pricing specific to that company.
That’s one of the other generalization techniques. Because AI problems in the enterprise generally have sparser datasets than if you’re trying to separate cat pictures from dog pictures. So, data sparcity is a constant challenge. I think transfer learning is one of the key strategies to avoid that.
You mentioned in passing, looking at things like games. I’ve often thought that was kind of a good litmus test for figuring out where to apply the technology, because games have points, and they have winners, and they have turns, and they have losers. They have structure to them. If that case study you just gave us was a game, what was the point in that? Was it a dollar of profit? Because you were like “Well, the plane could be, or it could fly here, where it might have a better chance to get somebody. But that’s got this cost. It wears out the plane, so the plane has to be depreciated accordingly.” What is the game it’s playing? How do you win the game it’s playing?
That’s a really great question. For XOJET, we actually created a tree of metrics, but at the top of the tree is something called fleet contribution, which is “What’s the profit generated per period of time, for the entire fleet?” Then, you can decompose that down to how many jets are flying, the length of haul, and the yield, which is the amount of dollars per hour flown. There’s also, obviously, a customer relationship component to it. You want to make sure that you get really good customers, and that you can serve them well. But there are very big differences between games and real-life business. Games have a finite number of moves. The rules are well-defined. There’s generally, if you look at Deep Blue or Alpha Go, or Arthur Samuels, or even the Labradas. All of these were two-player games. In the enterprise, you have typically tens, sometimes hundreds of players in the game, with undefined sets of moves. So, in the one sense, it’s a lot more complicated. The idea is, how do you reduce it, so it is game-like? That’s a very good question.
So, do you find that most people come to you with a defined business problem, and they’re not really even thinking about “I want some of this AI stuff. I just want my planes to be where they need to be.” What does that look like in the organization that brings people to you, or brings people to considering an artificial intelligence solution to a problem?
Typically, clients will see our success in one area, and then want to talk to us. For instance, we have a really great relationship with a steel company in Arkansas, called Big River Steel. Big River Steel, we’re building the world’s first learning steel mill with them. Which will learn from their sensors, and be able to just do all kinds of predictions and recommendations. It goes through that sense, propose, predict and score. It goes through that. So, when people heard that story, we got a lot of calls from steel mills. Now, we’re kind of deluged with calls from steel mills all over the world, saying, “How did you do that, and how do we get some of it?”
Typically, people hear about us because of AI. We’re a product company, with applications, so we generally don’t go in from a consulting point of view, and say “Hey, what’s your business problem?” We will generally go in and say, “Here are the ten areas where we have expertise and technology to improve business operations,” and then we’ll qualify a company, if it applies or not. One other thing is that AI follows the scientific methods, so it’s all about hypothesis, test, hypothesis, test. So it is possible that an AI application that works for one company will not work for another company. Sometimes, it’s the datasets. Sometimes, it’s just a different circumstance. So, I would encourage companies to be launching lots of hypotheses, using AI.
Your website has a statement quite prominently, “AI is not magic. It’s data.” While I wouldn’t dispute it, I’m curious. What were you hearing from people that caused you to… or maybe hypothetically, – you may not have been in on it – but what do you think is the source of that statement?
I think there’s a tremendous amount of hype and B.S. right now out there about AI. People anthropomorphize AI. You see robots with scary eyes, or you see crystal balls, or you see things that – it’s all magic. So, we’re trying to be explainers in chief, and to kind of de-mystify this, and basically say it’s just data and math, and supercomputers, and business expertise. It’s all of those four things, coming together.
We just happen to be at the right place in history, where there are breakthroughs in those areas. If you look at computing power, I would single that out as the thing that’s made a huge difference. In April of last year, NVIDIA released the DGX-1, which is their AI supercomputer. We have one of those in our data center, that in our platform we affectionately call “the beast,” which has a petaflop of computing power.
If you put that into perspective, that the fastest supercomputer in the world in the year 2000, was the ASCI Red, that had one teraflop of computing power. There was only one in the world, and no company in the world had access to that.
Now, with the supercomputing that’s out there, the beast has 1,000 times more computing power than the ASCI Red did. So, I think that’s a tremendous breakthrough. It’s not magic. It’s just good technology. The math behind artificial intelligence still relies largely on mathematical breakthroughs that happened in the ‘50s and ‘60s. And of course, Thomas Bayes, with Bayes’ Theorem, who was a philosopher in the 1700s.
There’s been a lot of good work recently around different variations on neural nets. We’re particularly interested in long- and short-term memory, and convolutional neural nets. But a lot of this is, a lot of the math has been around for a while. In fact, it’s why I don’t think we’re going to hit general intelligence any time soon. Because it is true that we have had exponential growth in computing power, and exponential growth in data. But it’s been a very linear growth in mathematics, right? If we start seeing AI algorithms coming up with breakthroughs in mathematics, that we simply don’t understand, then I think the antennas can go up.
So, if you have your DGX-1, at a petaflop, and in five years, you get something that’s an exaflop – it’s 1,000 times faster than that – could you actually put that to use? Or is it at some point, the jet company only has so much data. There are only so many different ways to crunch it. We don’t really need more – we have, at the moment, all of the processor power we need. Is that the case? Or would you still pay dearly to get a massively faster machine?
We could always use more computing power. Even with the DGX-1. For instance, we’re working with a distribution company where we’re generating 500,000 models a day for them, crunching on massive amounts of data. If you have massive datasets for your processing, it takes a while. I can tell you, life is a lot better. I mean, in the ‘90s, we were working on a neural net for the Coast Guard; to try to determine which ships off of the west coast were bad guys. It was very simple neural nets. You would hit return, and it would usually crash. It would run for days and days and days and days, be very, very expensive, and it just didn’t work.
Even if it came up with an answer, the ships were already gone. So, we could always use more computing power. I think right now, a limitation is more on the data side of it, and related to the fact that they shouldn’t be throwing out data that they’re throwing out. For instance, like customer relationship management systems. Typically, when you have an update to a customer, that it overwrites the old data. That is really, really important data. I think coming up with a proper data strategy, and understanding the value of data, is really, really important.
What do you think, on this theme of AI is not magic, it’s data; when you go into an organization, and you’re discussing their business problems with them, what do you think are some of the misconceptions you hear about AI, in general? You said it’s overhyped, and glowing-eyed robots and all of that. From an enterprise standpoint, what is it that you think people are often getting wrong?
I think there’s a couple of fundamental things that people are getting wrong. One is I think there is a tremendous over-reliance and over-focus on unstructured data, that people are falling in love with natural language processing, and thinking that that’s artificial intelligence. While it is true that NLP can help with judging things like consumer sentiment or customer feedback, or trend analysis on social media, generally those are pretty weak signals. I would say, don’t follow the shiny object. I think the reason people see that, is the success of Siri and Alexa, and people see that as AI. It is true that those are learning algorithms, and those are effective in certain circumstances.
I think they’re much less effective when you start getting into dialogue. Doing dialogue management with humans is extraordinarily difficult. Training the corpus of those systems is very, very difficult. So, I would say stay away from chatbots, and focus mostly on structured data, rather than unstructured data. I think that’s a really big one. I also think that focusing on the supply side of a company is actually a much more fruitful area than focusing on the demand side, other than sales forecasting. The reason I say that is that the interactions between inbound materials and production, and distribution, are more easily modeled and can actually make a much bigger difference. It’s much harder to model things like the effect of a promotion on demand, although it’s possible to do a lot better than they’re doing now. Or, things like customer loyalty; like the effect of general advertising on customer loyalty. I think those are probably two of the big areas.
When you see large companies being kind of serious about machine learning initiatives, how are they structuring those in the organization? Is there an AI department, or is it in IT? Who kind of “owns” it? How are its resources allocated? Are there a set of best practices, that you’ve gleaned from it?
Yes. I would say there are different levels of maturity. Obviously, the vast majority of companies have no organization around this, and it is individuals taking initiatives, and experimenting by themselves. IT in general has not taken a leadership role in this area. I think, fundamentally, that’s because IT departments are poorly designed. Like the CIO job needs to be two jobs. There needs to be a Chief Infrastructure Officer and Chief Innovation Officer. One of those jobs is to make sure that the networks are working, the data center is working, and people have computers. The other job is, “How are advances in technologies helping companies?” There are some companies that have Chief Data Officers. I think that’s also caused a problem, because they’re focusing more on big data, and less on what do you actually do with those data?
I think the most advanced companies – I would say, first of all, it’s interesting, because it’s following the same trajectory as information technology organizations follow, in companies. First, it’s kind of anarchy. Then, there’s the centralized group. Then, it goes to a distributed group. Then, it goes to a federated group, federated meaning there’s a central authority which basically sets standards and direction. But each individual business unit has their representatives. So, I think we’re going to go through a whole bunch of gyrations in companies, until we end up where most technology organizations are today, which is; there is a centralized IT function, but each business unit also has IT people in it. I think that’s where we’re going.
And then, the last question along these lines: Do you feel that either: A) machine learning is doing such remarkable things, and it’s only going to gain speed, and grow from here, or B) machine learning is over-hyped to a degree that there are unrealistic expectations, and when disappointment sets in, you’re going to get a little mini AI winter again. Which one of those has more truth?
Certainly, there is a lot of hype about it. But I think if you look at the reality of how many companies have actually implemented learning algorithms; AI, ML, data science, across the operations of their company, we’re at the very, very beginning. If you look at it as a sigmoid, or an s-curve, we’re just approaching the first inflection point. I don’t know of any company that has fully deployed AI across all parts of their operations. I think ultimately, executives in the 21stcentury will have many, many learning algorithms to support them, making complex business decisions.
I think the company that clearly has exhibited the strongest commitment to this, and is furthest along, is Amazon. If you wonder how Amazon can deliver something to your door in one hour, it’s because there are probably 100 learning algorithms that made that happen, like where should the distribution center be? What should be in the distribution center? Which customers are likely to order what? How many drivers do we need? What’s the route the driver should take? All of those things are powered by learning algorithms. And you see the difference, you feel the difference, in a company that has deployed learning algorithms. I also think if you look back, from a societal point of view, that if we’re going to have ten billion people on the planet, we had better get a lot more efficient at the consumption of natural resources. We had better get a lot more efficient at production.
I think that means moving away from static business rules that were written years ago, that are only marginally relevant to learning algorithms that are constantly optimizing. And then, we’ll have a chance to get rid of what Hackett Group says is an extra trillion dollars of working capital, basically inventory, sitting in companies. And we’ll be able to serve customers better.
You seem like a measured person, not prone to wild exaggeration. So, let me run a question by you. If you had asked people in 1995, if you had said this, “Hey, you know what? If you take a bunch of computers, just PCs, like everybody has, and you connected them together, and you got them to communicate with hypertext protocol of some kind, that’s going to create trillions and trillions and trillions and trillions and trillions of dollars of wealth.” “It’s going to create Amazon and Google and Uber and eBay and Etsy and Baidu and Alibaba, and millions of jobs that nobody could have ever imagined. And thousands of companies. All of that, just because we’re snapping together a bunch of computers in a way that lets them talk to each other.” That would have seemed preposterous. So, I ask you the question; is artificial intelligence, even in the form that you believe is very real, and what you were just talking about, is it an order of magnitude bigger than that? Or is it that big, again? Or is it like “Oh, no. Just snapping together, a bunch of computers, pales to what we are about to do.” How would you put your anticipated return on this technology, compared to the asymmetrical impact that this seemingly very simple thing had on the world?
I don’t know. It’s really hard to say. I know it’s going to be huge. Right? It is fundamentally going to make companies much more efficient. It’s going to allow them to serve their customers better. It’s going to help them develop better products. It’s going to feel a lot like Amazon, today, is going to be the baseline of tomorrow. And there’s going to be a lot of companies that – I mean, we run into a lot of companies right now that just simply resist it. They’re going to go away. The shareholders will not tolerate companies that are not performing up to competitive standards.
The competitive standards are going to accelerate dramatically, so you’re going to have companies that can do more with less, and it’s going to fundamentally transform business. You’ll be able to anticipate customer needs. You’ll be able to say, “Where should the products be? What kind of products should they be? What’s the right product for the right customer? What’s the right price? What’s the right inventory level? How do we make sure that we don’t have warehouses full of billions and billions of dollars worth of inventory?”
It’s very exciting. I think the business, and I’m generally really bad at guessing years, but I know it’s happening now, and I know we’re at the beginning. I know it’s accelerating. If you forced me to guess, I would say, “10 years from now, Amazon of today will be the baseline.” It might even be shorter than that. If you’re not deploying hundreds of algorithms across your company, that are constantly optimizing your operations, then you’re going to be trailing behind everybody, and you might be out of business.
And yet my hypothetical 200-person company shouldn’t do anything today. When is the technology going to be accessible enough that it’s sort of in everything? It’s in their copier, and it’s in their routing software. When is it going to filter down, so that it really permeates kind of everything in business?
The 200-person company will use AI, but it will be in things like, I think database design will change fundamentally. There is some exciting research right now, actually using predictive algorithms to fundamentally redesign database structures, so that you’re not actually searching the entire database; you’re just searching most likely things first. Companies will use AI-enabled databases, they’ll use AI in navigation, they’ll use AI in route optimization. They’ll do things like that. But when it comes down to it, for it to be a good candidate for AI, in helping make complex decisions, the answer needs to be non-obvious. Generally with a 200-person company, having run a company that went from 2 people to 20 people, to 200 people, to 2,000 people, to 20,000 people, I’ve seen all of the stages.
A 200-person company, you can kind of brute force. You know everybody. You’ve just crossed Dunbar’s number, so you kind of know everything that’s going on, and you have a good feel for things. But like you said, I think applying it in using other peoples’ technologies that are driven by AI, for the things that I talked about, will probably apply to a 200-person company.
With your jet company, you did a project, and EBITDA went up 5%, and that was a big win. That was just one business problem you were working on. You weren’t working on where they buy jet fuel, or where they print. Nothing like that. So presumably, over the long haul, the technology could be applied in that organization, in a number of different ways. If we have a $70 trillion economy in the world, what percent is – 5% is easy – what percentage improvement do you think we’re looking at? Like just growing that economy dramatically, just by the efficiencies that machine learning can provide?
Wow. The way to do that is to look at an individual company, and then sort of extrapolate. I would say an individual company could, if you look at the value of companies. That’s the way I look at it, like shareholder value, which is made up of revenue, margins and capital efficiency. I think that revenue growth could take off, could probably double, from what it is. The growth could double from what it is now. Margins, it will have a dramatic impact. I think you could, if you look at all of the different things you could do within the company, and you had fully deployed learning algorithms, and gotten away from making decisions on yardsticks and averages, you could, a typical company, I’ll say double your margins.
But the home run is in capital efficiency, which not too many people pay attention to, and is one of the key drivers of return on invested capital, which is the driver of general value. This is where you can reduce things 30%, things like that, and get rid of warehouses of stuff. That allows you to be a lot more innovative, because then you don’t have obsolescence. You don’t have to push products that don’t work. You can develop more innovative products. There are a lot of good benefits. Then, you start compounding that year over year, and pretty soon, you’ve made a big difference.
Right, because doubling margins alone doubles the value of all of the companies, right?
It would, if you projected it out over time. Yes. All else being equal.
Which it seldom is. It’s funny, you mentioned Amazon earlier. I just assumed they had a truck with a bunch of stuff on it, that kept circling my house, because it’s like every time I want something, they’re just there, knocking on the door. I thought it was just me!
Yeah. Amazon Prime now came out, was it last year? In the Bay Area? My daughter ordered a pint of ice cream and a tiara. An hour later, a guy is standing at the front door with a pint of ice cream, and a tiara. It’s like Wow!
What a brave new world, that has such wonders in it!
Exactly!
As we’re closing up on time here, there are a number of people that are concerned about this technology. Not in the killer robot scenario. They’re concerned about automation; they’re concerned about – you know it all. Would you say that all of this technology and all of this growth, and all of that, is good for workers and jobs? Or it’s bad, or it’s disruptive in the short term, not in the long term? How do you size that up for somebody who is concerned about their job?
First of all, moving sort of big picture to small picture, first of all, this is necessary for society, unless we stop having babies. We need to do this, because we have finite resources, and we need to figure out how to do more with less. I think the impact on jobs will be profound. I think it will make a lot of jobs a lot better. In AI, we say it’s augment, amplify and automate. Right now, like the things we’re doing at XOJET really help make the people in revenue management a lot more powerful, and I think, enjoy their jobs a lot more, and doing a lot less routine research and grunt work. So, they actually become more powerful, it’s like they have super powers.
I think that there will also be a lot of automation. There are some tasks that AI will just automate, and just do, without human interaction. A lot of decisions, in fact most decisions, are better if they’re made with an algorithm anda human, to bring out the best of both. I do think there’s going to be a lot of dislocation. I think it’s going to be very similar to what happened in the automotive industry, and you’re going to have pockets of dislocation that are going to cause issues. Obviously, the one that’s talked about the most is the driverless car. If you look at all of the truck drivers, I think probably within a decade, that most cross-country trucks, there’s going to be some person sitting in their house, in their pajamas, with nine screens in front of them, and they’re going to be driving nine trucks simultaneously, just monitoring them. And that’s the number one job of adult males in the U.S. So, we’re going to have a lot of displacement. I think we need to take that very seriously, and get ahead of it, as opposed to chasing it, this time. But I think overall, this is also going to create a lot more jobs, because it’s going to make more successful companies. Successful companies hire people and expand, and I think there are going to be better jobs.
You’re saying it all eventually comes out in the wash; that we’re going to have more, better jobs, and a bigger economy, and that’s broadly good for everyone. But there are going to bumps in the road, along the way. Is that what I’m getting from you?
Yes. I think it will actually be a net positive. I think it will be a net significant positive. But it is a little bit of, as economists would say, “creative destruction.” As you go from agricultural to industrial, to knowledge workers, toward sort of an analytics-driven economy, there are always massive disruptions. I think one of the things that we really need to focus on is education, and also on trade schools. There is going to be a lot larger need for plumbers and carpenters and those kinds of things. Also, if I were to recommend what someone should study in school, I would say study mathematics. That’s going to be the core of the breakthroughs, in the future.
That’s interesting. Mark Cuban was asked that question, also. He says the first trillionaires are going to be in AI.  And he said philosophy. Because in the end, what you’re going to need are what the people know how to do. Only people can impute value, and only people can do all of that.
Wow! I would also say behavioral economics; understanding what humans are good at doing, and what humans are not good at doing. We’re big fans of Kahneman and Tversky, and more recently, Thaler. When it comes down to how humans make decisions, and understanding what skills humans have, and what skills algorithms have, it’s very important to understand that, and to optimize that over time.
All right. That sounds like a good place to leave it. I want to thank you so much for a wide-ranging show, with a lot of practical stuff, and a lot of excitement about the future. Thanks for being on the show.
My pleasure. I enjoyed it. Thanks, Byron.
Byron explores issues around artificial intelligence and conscious computers in his new book The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity.
[voices_in_ai_link_back]

Voices in AI – Episode 43: A Conversation with Markus Noga

[voices_in_ai_byline]
In this episode, Byron and Markus discuss machine learning and automation.
[podcast_player name=”Episode 43: A Conversation with Markus Noga” artist=”Byron Reese” album=”Voices in AI” url=”https://voicesinai.s3.amazonaws.com/2018-05-22-(00-58-23)-markus-noga.mp3″ cover_art_url=”https://voicesinai.com/wp-content/uploads/2018/05/voices-headshot-card.jpg”]
[voices_in_ai_byline]
Byron Reese: This is Voices In AI brought to you by GigaOm, I’m Byron Reese. Today, my guest is Marcus Noga. He’s the VP of Machine Learning over at SAP. He holds a Ph.D.in computer science from Karlsruhe Institute of Technology, and prior to that spent seven years over at Booz Allen Hamilton working on helping businesses adopt and transform their businesses through IT. Welcome to the show Markus.
Markus Noga: Thank you Byron and it’s a pleasure to be here today.
Let’s start off with a question I have yet to have two people answer the same way. What is artificial intelligence?
That’s a great one, and it’s sure something that few people can agree on. I think the textbook definition mostly defines that by analogy with human intelligence, and human intelligence is also notoriously tricky and hard to define. I define human intelligence as the ability to deal with the unknown and bring structure to the unstructured, and answer novel questions in a surprisingly resourceful and mindful way. Artificial intelligence in itself is the thing, rather more playfully, that is always three to five years out of reach. We love to focus on what can be done today—what we call machine learning and deep learning—that can draw a tremendous value for businesses and for individuals already today.
But, in what sense is it artificial? Is it artificial intelligence in the way artificial turf? Is it really turf, it just looks like it? Or is it just artificial in the sense that we made it? Or put another way, is artificial intelligence actually intelligent? Or does is it just behave intelligently?
You’re going very deep here into things like Searle’s Chinese room paradox about the guy in the room with a hand for definitions of how to transcribe Chinese symbols to have an intelligent conversation. The question being who or what is having the intelligent conversation. Is it the book? Certainly not. Is it the guy mindlessly transcribing these symbols? Certainly not? Is it maybe the system of the guy in the room, the book, and the room itself that generates these intelligent seeming responses? I guess I’m coming down on the output-oriented side here. I try not to think too hard about the inner states or qualia, or the question whether the neural networks we’re building have a sentient experience or the experience in this qualia. For me, what counts is whether we can solve real-world problems in a way that’s compatible with intelligence. Its place in intelligent behavior of everything else—I would leave to the philosophers Byron.
We’ll get to that part where we can talk about the effects of automation and what we can expect and all of that. But, don’t you think at some level, understanding that question, doesn’t it to some degree inform you as to what’s possible? What kinds of problems should we point this technology at? Or do you think it’s entirely academic that it has no real-world implications?
I think it’s extremely profound and it could unlock a whole new curve of value creation. It’s also something that, in dealing with real-world problems today, we may not have to answer—and this is maybe also something specific to our approach. You’ve seen all these studies that say that X percent of activities can be automated with today’s machine learning, and Y percent could be automated if there are better natural language speech processing capabilities and so on, and so forth. There’s such tremendous value to be had by going after all these low-hanging fruits and sort of doing applied engineering by bringing ML and deep learning into an application context. Then we can bide our time until there is a full answer to strong AI, and some of the deeper philosophical questions. But what is available now is already delivering tremendous value, and will continue to do so over the next three to five years. That’s my business hat on—what I focus on together with the teams that I’m working with. The other question is one that I find tremendously interesting for my weekend and unique conversations.
Let me ask you a different one. You started off by saying artificial intelligence, and you dealt with that in terms of human intelligence. When you’re thinking of a problem that you’re going to try to use machine intelligence to solve, are you inspired in any way by how the brain works or is that just a completely different way of doing it? Or do we learn how intelligence, with the capital I, works by studying the brain?
I think that’s the multi-level answer because clearly the architectures that do really well in analytic learning today are in a large degree neurally-inspired. Instead of having multi-layered deep networks—having them with a local connection structure, having them with these things we call convolutions that people use in computer vision, so successfully—it resembles closely some of the structures that you see in the visual cortex with vertical columns for example. There’s a strong argument for both these structures in the self-referential recurrent networks that people use a lot for video processing and text processing these days are very, very deeply morally inspired. On the other hand, we’re also seeing that a lot of the approaches that make ML very successful today are about as far from neutrally-inspired learning as you can get.
Example one, we struggled as a discipline with neutrally-inspired transfer functions—that were all nice, and biological, and smooth—and we couldn’t really train deep networks with them because they would saturate. One of the key enablers for modern deep learning was to step away from the biological analogy of smooth signals and go to something like the rectified linear unit, the ReLU function, as an activation, and that has been a key part in being able to train very deep networks. Another example when a human learns or an animal learns, we don’t tend to give them 15 million cleanly labored training examples, and expect them to go over these training examples 10times in a row to arrive at something. We’re much closer to one-shot learning and being able to recognize the person with a cylinder hat on their head just the basis of one description or one image that shows us something similar.
So clearly, the approaches that are most successful today are both sharing some deep neural inspiration as a basis, but, also a departure into computationally tractable, and very, very different kinds of implementations than the network that we see in our brains. I think that both of these themes are important in advancing the state-of-the-art in ML and there’s a lot going on. In areas like one-shot learning, for example, right now I’m trying to mimic more of the way the human brain—with an active working memory and these rich associations—is able to process new information, and there’s almost no resemblance to what convolutional networks and the current networks do today.
Let’s go with that example. If you take a small statue of a falcon, and you put it in a hundred photos—and sometimes it’s upside down, and sometimes it’s laying on its side, sometimes it’s half in water, sometimes it’s obscured, sometimes it’s in shadows—a person just goes “boom boom boom boom boom” and picks them out, right and left with no effort, you know, one-shot learning. What do you think a human is doing? It is an instance of some kind of transfer learning, but what do you think is really going on in the human brain, and how do you map that to computers? How do you deal with that?
This is an invitation to speculate on the topics of falcons, so let me try. I think that, clearly, our brains have built a representation of the real world around us, because we’re able to create that representation even though the visual and other sensory stimuli that reach us are not in fact as continuous as they seem. Standing in the room here having the conversation with you, my mind creates the illusion of a continuous space around me, but in fact, I’m getting distinct feedbacks from the eyes as they succumb and jump around the room. The illusion of a continuous presence, the continuous sharp resolution of the room is just that; it’s an illusion because our mind has built very, very effective mental models of the world around us, that’s highly contrasting information and make it tractable on an abstract level.
Some of the things that are going on in research right now [are] trying to exploit these notions, and trying to use a lot of unsupervised training with some very simple assumptions behind them; basically the mind doesn’t like to be surprised, and would, therefore, like to predict what’s next [by]leveraging very, very powerful unsupervised training approaches where you can use any kind of data that’s available, and you don’t need to enable it to come up with these unsupervised representation learning approaches. They seem to be very successful, and they’re beating a lot of the traditional approaches because you can have access to way larger corpuses of unlabeled information which means you can train better models.
Now is that it a direct analogy to what the human brain does? I don’t know. But certainly it’s an engineering strategy that results in world-leading performance on a number of very popular benchmarks right now, and it is, broadly speaking, neutrally-inspired. So, I guess bringing together what our brains do and what we can do in engineering is always a dance between the abstract inspiration that we can get from how biology works, and the very hard math and engineering in getting solutions to train on large-scale computers with hundreds of teraflops in compute capacity and large matrix multiplications in the middle. It’s advances on both sides of the house that make ML advance rapidly today.
Then take a similar problem, or tell me if this is a similar problem, when you’re doing voice recognition, and there’s somebody outside with the jackhammer, you know, it’s annoying, but a human can separate those two things. It can hear what you’re saying just fine, but for a machine, that’s a really difficult challenge. Now my question to you is, is that the same problem? Is it one trick humans have like that that we apply in a number of ways? Or is that a completely different thing that’s going on in that example?
I think it’s similar, and you’re hitting onto something because in the listening example there are some active and some passive components going on. We’re all familiar with the phenomenon of selective hearing when we’re at a dinner party, and there are 200 conversations going on in parallel. If we focus our attention on a certain speaker or a certain part of the conversation, we can make them stand out over the din and the noise because their own mind had some prior assumptions as to what constitutes a conversation, and we can exploit these priors in our minds in order to selectively listen in to parts of the conversation. This has partly a physical characteristic, maybe hearing in stereo. Our ears have certain directional characteristics to the way they pick up certain frequencies by turning our head the right way and inclining it the right way. We can do a lot already [with] stereo separation, whereas, if you have a single microphone—and that’s all the signal you get—all these avenues would be closed to you.
But, I think the main story is one about signals superimposed with noise—whether that’s camera distortions, or fog, or poor lighting in the case of the statue that we are trying to recognize, or whether it’s ambient noise or intermittent outages in the sense of the audio signal that you’re looking into. The two different most popular neutrally-inspired architectures on the market right now, [are] the convolutional networks for a lot of things in the image and also natural text space, and the recurrent networks for a lot of things in the audio ends at time series signal, but also on text space. Both share the characteristics that they are vastly more resilient to noise than any hard-coded or programmed approach. I guess the underlying problem is one that, five years ago, would have been considered probably unsolvable; where today with these modern techniques, we’re able to train models that can adequately deal with the challenges if the information is in the solid state.
Well, what do you think when the human hears, at a conversation at the party to go with that example, and you kind of like, “Oh, I want to listen to that.” I heard what you say that there’s one aspect of you where you make a physical modification to the situation, but what you’ve also done is introduced this idea of consciousness, that a person selectively can change their focus and that aspect of what the brain is doing, where it’s like, “Oh, wait a minute.” Maybe something that’s hard to implement on a machine, or is that not the case at all?
If you take that idea, and I think in the ML research and engineering communities this is currently most popular under the label of attention, or attention-based mechanisms, then certainly this is all over leading approaches right now—whether it’s the computer vision papers from CVPR just last week or whether it’s the text processing architectures that return state-of-the-art results right now. They all start to include some kind of attention mechanism allowing you to both weigh outputs by the center of attention, and also to trace back results to centers of extension, which have two very nice properties. On the one hand attention mechanisms, nascent as they are today, help improve the accuracy of what models can deliver. On the second hand, the ability to trace back on the outcome of a machine learning model to centers and regions of attention in the input can do wonders for explain-ability of ML and AI results, which is something that increasingly users and customers are looking for. Don’t just give me any result which is as good as my current process, or hopefully a couple of percentage points better. But, also helped me build confidence in this by explaining why things are being classed or categorized or translated or extracted the way they are. To gain the human trust into operating system of humans and machines working together explain-ability future is big.
One of the peculiar things to me, with regard to strong AI—general intelligence—is that there are folks who say, when you say, “When will we get a general intelligence, “the soonest you ever hear is five years. There are very famous people who believe we’re going to have something very soon. Then you get the other extreme is about 500 years and that worrying about that is like worrying about overpopulation on Mars. My question to you is why do you think that there’s such a wide range in terms of our idea of when we may make such a breakthrough?
I think it’s because of one vexing property of humans and machines is that the things that are easiest for us humans tend to be the things that are hardest for machines and vice versa. If you look at that today, nobody would dream of having computer as a job description. That’s a machine. If you think back 60-70 years, computer was the job description of people actually doing manual calculations. “Printer” was a job description, and a lot of other things that we would never dream of doing manually today were being done manually. Think of spreadsheets potentially the greatest simple invention in computing, think of databases, think of things like enterprise resource planning systems that SAP does, and business networks connecting them or any kind of cloud-based solutions—what they deliver is tremendous and it’s very easy for machines to do, but it tends to be the things that are very hard for humans. Now at the same time things that are very easy for humans to do, see a doggie and shout “doggie,” or see a cat and say “meow” is something that toddlers can do, but until very, very recently, the best and most sophisticated algorithms haven’t been able to do that part.
I think part of the excitement around ML and deep learning right now is that a lot of these things have fallen, and we’re seeing superhuman performance on image classification tasks. We’re seeing superhuman performance on things like switchboard voice-to-text transcription tasks, and many other elements are falling to machines that that used to be very easy for humans but are now impossible for us. This is something that generates a lot of excitement right now. I think where we have to be careful is [letting] this guide our expectations on the speed of progress in following years. Human intuition about what is easy and what is hard is traditionally a very, very poor guide to the ease of implementation with computers and with ML.
Example, my son was asking me yesterday, “Dad, how come the car can know where it is at and tell us where to drive?” And I was like, “Son, that’s fairly straightforward. There are all these satellites flying around, and they’re shouting at us, ‘It’s currently 2 o’clock and 30 seconds,’ and we’re just measuring the time between their shouts to figure out where we are today, and then that gives us that position on the planet. It’s not a great invention; it’s the GPS system—it’s mathematically super hard to do for a human with a slide rule; it’s very easy to do for the machine.” And my son said, “Yeah, but that’s not what I wanted to know. How come the machine is talking to us with the human voice? This is what I find amazing, and I would like to understand how that is built.” and I think that our intuition about what’s easy and what’s hard is historically a very poor guide for figuring out what the next step and the future of ML and artificial intelligence look like. This is why you’re getting those very broad bands of predictions.
Well do you think that the difference between the narrow or weak AI we have now and strong AI, is evolutionary? Are we on the path [where] when machines get somewhat faster, and we get more data, and we get better algorithms, that we’re going to gradually get a general intelligence? Or is a general intelligence something very different, like a whole different problem than the kinds of problems we’re working on today?
That’s a tough one. I think that taking the brain analogy; we’re today doing the equivalent of very simple sensory circuits which maybe can’t duplicate the first couple of dozens or maybe a hundred layers in the way the visual cortex works. We’re starting to make progress into some things like one-shot learning; it’s very nascent in that early-stage research right now. We’re starting to make much more progress in directions like reinforcement learning, but overall it’s very hard to say which if any additional mechanisms are there in the large. If you look at the biological system of the brain, there’s a molecular level that’s interesting. There’s a cellular level that’s interesting. There is a simple interconnection I know that’s interesting. There is a micro-interconnection level that’s interesting. I think we’re still far from a complete understanding of how the brain works. I think right now we have tremendous momentum and a very exciting trajectory with what our artificial neural networks can do, and at least for the next three to five years. There seems to be pretty much limitless potential to bring them out into real-world businesses, into real-world situations and contexts, and to create amazing new solutions. Do I think that really will deliver strong AI? I don’t know. I’m an agnostic, so I always fall back to the position that I don’t know enough.
Only one more question about strong AI and then let’s talk about the shorter-term future. The question is, human DNA converted to code is something like 700 MB, give or take. But the amount that’s uniquely human, compared to say a chimp or something like that is only about 1% difference—only 7 or 8 or 9 MB of code—is what gives us a general intelligence. Does that imply or at least tell us how to build something that then can become generally intelligent? Does that imply to you that general intelligence is actually simple, straightforward? That we can look at nature and say, it’s really a small amount of code, and therefore we really should be looking for simple, elegant solutions to general intelligence? Or do those two things just not map at all?
Certainly, what we’re seeing today is that deep learning approaches to problems like image classification, image object detection, image segmentation, video annotation, audio transcription—all these things tend to be orders of magnitude, smaller problems than what we dealt with when we handcrafted things. The core of most deep learning solutions to these things, if you really look at the core model on the model structure, tends to be maybe 500 lines of code, maybe 1000. And that’s within the reach of an individual putting this together over a weekend, so the huge democratization that deep learning based on big data lends is that actually a lot of these models that do amazing things are very, very small code artifacts. The weight matrices and the binary models that they generate then tend to be as large or larger than traditional programs compiled into executable, sometimes orders of magnitude larger again. The thing is, they are very hard to interpret, and we’re only at the beginning of an explain-ability of what the different weights and the different excitations mean. I think there are some nice early visualizations on this. There are also some nice visualizations that explain what’s going on with attention mechanisms in the artificial networks.
As to explain-ability of the real network in the brain, I think that is very nascent. I’ve seen some great papers and results on things like spatial representations in the visual cortex where surprisingly you find triangle scripts or attempts to reconstruct the image hitting the retina based on reading, with fMRI scans, the excitations in lower levels of the visual cortex. They show that we’re getting closer to understanding the first few layers. I think that even with the 7 MB difference or so that you allude to between chimps and humans spelled out for us, there is a whole set of layers of abstractions between the DNA code and the RNA representation, the protein representation, the excitation of these with methylation and other mechanisms that control activation of genes, and the interplay of the proteins across a living breathing human brain that all of this magnitude of complexity above of the super megabyte, by a certain megabyte difference in A’s and C’s, and T’s, and G’s. We live in super exciting types. We live in times were a new record, and a new development, and a new capability that was unthinkable of a year ago, or let alone a decade ago, is becoming commonplace, and it’s an invigorating and exciting time to be alive. I still struggle to make a prediction from the year to general AI based on a straight-line trend.
There’s some fear wrapped up though as exciting as AI is, there’s some fear wrapped up in it as well. The fear is the effect of automation on employment. I mean you know this, of course, it’s covered so much. There’s kind of three schools of thought: One says that we’re going to automate certain tasks and that there will be a group of individuals who do not have the training to add economic value. They will be pushed out of the labor market, and we’ll have perpetual unemployment, like a big depression that never goes away. Then there’s another group that says, “No, no, no, you don’t understand. Everybody is replaceable. Every single job we have, machines can do any of it.” And then there’s a third school about that says, “No, none of that’s going to happen. The history of 250 years of the Industrial Revolution is that people take these new technologies, even profound ones like electricity and engines, and steam, and they just use them to increase their own productivity and to drive wages up. We’re not going to have any unemployment from this, any permanent unemployment.” Which of those three camps, or a fourth, do you fall into?
I think that there’s a lot of historical precedent for how technology gets adopted, and there are also numbers of the adoption of technologies in our own day and age that sort of serve as reference points here. For example, one of the things that surprised me, truly, is the amount of e-commerce—as a percentage of overall retail market share—[that] is still in the mid to high single digit percentage points according to surveys that I’ve seen. That totally does not match my personal experience of basically doing all my non-grocery shopping entirely online. But it shows that in the 20-25 years of the Internet Revolution, a tremendous value has been created—and the conveniences of having all kinds of stuff at your doorstep with just a single click actually—that has transformed the single-digit percentage of the overall retail market with the transformation that we’ve seen. This was one of the most rapid uptakes in history of new technology that has groundbreaking value, by decoupling evidence and bits, and it’s been playing out over the past 20-25 years that all of us are observing.
So, I think while there is tremendous potential of machine learning in AI to drive another Industrial Revolution, we’re also in the middle of all these curves from other revolutions that are ongoing. We’ve had a mobile revolution that unshackled computers and gave everybody what used to be a supercomputer in their pocket which had an infinite revolution. Before that, we’ve had a client-server revolution and the computing revolution in its own—all of these building on prior revolutions like electricity, or the internal combustion engine, or methods like the printing press. They certainly have a tendency to show accelerating technology cycles. But on the other hand, for something like e-commerce or even mobile, the actual adoption speed has been one that is none too frightening. So for all the tremendous potential that ML and AI bring, I would be hard-pressed to come up with a completely disruptive scenario here. I think we are seeinga technology with tremendous potential for rapid adoption. We’re seeing the potential to both create new value and do new things, and to automate existing activities which continues past trends. Nobody has computer or printer as their job description today, and job descriptions like social-media influencer, or blogger, or web designer did not exist 25 years ago. This is an evolution on a Schumpeterian creative destruction that is going on all over industry, in every industry, in every geography, based on every new technology curve that comes in here.
I would say fears in this space are greatly overblown today. But fear is real the moment you feel it, therefore institutions—like The Partnership on Artificial Intelligence, with the leading technology companies, as well as the leading NGOs, think tanks, and research institutes—are coming together to discuss the implications of AI, and ethics of AI, and safety and guiding principles. All of these things are tremendously important to make sure that we can adopt this technology with confidence. Just remember that when cars were new, Great Britain had a law that a person with a red flag had to walk in front of the car in order to warn all pedestrians of the danger that was approaching. That was certainly an instance of fear about technology, that, on the one hand, was real at that point in time, but that also went away with a better understanding of how it works and of the tremendous value on the economy.
What do you think of these efforts to require that when an artificial intelligence makes a ruling or a decision about you that you have a right to know why it made that decision? Is that a manifestation of the red flag in front of the car as well, and is that something that would, if that became the norm, actually constrain the development of artificial intelligence?
I think you’re referring to the implicit right to explanation on this part of the European Union privacy novella for 2018. Let me start by saying that the privacy novella we’re seeing is a tremendous step forward because the simple act of harmonizing the rules and creating one digital playing field across the hundreds of millions of European citizens, and countries, and nationalities, is a tremendous step forward. We used to have one different data protection regime for each federal state in Germany, so anything that is required and harmonized is a huge step forward. I also think that the quest for an explanation is something that is very human. At the core of us is to continue to ask “why” and “how.” That is something that is innate to ourselves when we apply for a job with the company, and we get rejected. We want to know why. And when we apply for a mortgage and we can offer a rate that seems high to us and we want to understand why. That’s a natural question, it’s a human question, and it’s an information need that needs to be served if we don’t want to end up in a Kafka-esque future where people don’t have a say about their destiny. Certainly, that is hugely important on the one hand.
On the other hand, we also need to be sure that we don’t measure ML and AI to a stricter standard than we measure humans today because that could become an inhibitor to innovation. So, if you ask a company, “Why you didn’t get accepted for that job offer?” They will probably say, “Dear Sir or Madam, thank you for your letter. Due to the unusually strong field of candidates for this particular posting, we regret to inform you that certain others are stronger, and we wish you all the best for your continued professional future.” This is what almost every rejection letter reads like today. Are we asking the same kind of explain-ability from an AI system that is delivering a recommendation today that we apply to a system of humans and computers working together to create a letter like that? Or are we holding them to a much, much higher standard? If it is the first thing, absolutely essential. If it’s the second thing, we got to watch whether we’re throwing out the baby with the bathwater on this one. This is something where we, I think, need to work together to find the appropriate levels and standards for things like explain-ability in AI to fill very abstract sentences like write to an explanation with life that can be implemented, that can be delivered, and that can provide satisfactory answers at the same time while not unduly inhibiting progress. This is something that, with a lot of players focused on explain-ability today, where we will certainly see significant advances going forward.
If you’re a business owner, and you read all of this stuff about artificial intelligence, and neural nets, and machine learning, and you say, “I want to apply some of this great technology in my company,” how do people spot problems in a business that might be good candidates for an AI solution?
I can extort that and turn it around by asking, “What’s keeping you awake at night? What are the three big things that make you worried? What are the things that make up the largest part of your uncertainty, or of your cost structure, or of the value that you’re trying to create?” Looking on end-to-end processes, it’s usually fairly straightforward to identify cases where AI and ML might be able to help and to deliver tremendous value. The use-case identification tends to be the fairly easiest chord of the game. Where it gets tricky is in selecting and prioritizing these cases, figuring out the right things to build, and finding the data that you need in order to make the solution real, because unlike traditional software engineering, this is about learning from data. Without data, you basically can’t sort or at least we have to build some very small simulators in order to create the data that you’re looking for.
You mentioned that that’s the beginning of the game, but what makes the news all the time is when AI beats a person at a game. In 1997 you had chess, then you had Ken Jennings in Jeopardy!, then you had AlphaGo and Lee Sedol, and you had AI beating poker. Is that a valid approach to say, “Look around your business and look for things that look like games?” Because games have constrained rules, and they have points, and winners, and losers. Is that a useful way to think about it? Or are the game things more like AI’s publicity, a PR campaign, and that’s not really a useful metaphor for business problems?
I think that these very publicized showcases are extremely important to raise awareness and to demonstrate stunning new capabilities. What we see in building business solutions is that I don’t necessarily have to be the human world champion in something in order to deliver value. Because a lot of business is about processes, is about people following flowcharts together with software systems trying to deliver a repeatable process for things like customer service, or IT incident handling, or incoming invoice screening and matching, or other repetitive recurring tasks in the enterprise. And already by addressing—it’d be easy to serve 60-80% of these, we can create tremendous value for enterprises by making processes run faster, by making people more productive, and by relieving them of the parts of activities that they regard as repetitive and mind-numbing, and not particularly enjoyable.
The good thing is that in a modern enterprise today, people tend to have IT systems in place where all these activities leave a digital exhaust stream of data, and locking into that digital exhaust stream and learning from it is the key way to make ML solutions for the enterprise feasible today. This is one of the things where I’m really proud to be working for SAP because 76% of all business transactions, as measured by value, anywhere on the globe, are on an SAP system today. So if you want to learn models on digital information that touch the enterprise, chances are it’s either in an SAP system or in a surrounding system already today. Looking for these and sort of doing the intersection between what’s attractive—because I can serve core business processes with faster speed, greater agility, lower cost, more flexibility, or bigger value—and crossing that with the feasibility aspect of “do I have the digital information that I can learn from to build business-relevant functionality today?,” is our overriding approach to identifying things that we built in order to make all our SAP enterprise applications intelligent.
Let’s talk about that for a minute. What sorts of things are you working on right now? What sorts of things have the organization’s attention in machine learning?
It’s really end-to-end digital intelligence on processes, and let me give you an example. If you look at the finance space, which SAP is well-known for, these huge end-to-end processes—like record to report, or things like invoice to record—which really deal end-to-end with what an enterprise needs to do in order to buy stuff and pay for it, and receive it, or to sell stuff, and get paid for it. These are huge machines with dozens and dozens of process steps, and many individuals in shared service environments that otherwise perform the delivering of these services. They see a document like an invoice, for example, it’s just the tip of the iceberg for a complex orchestration and things to deal with that. We’re taking these end-to-end processes, and we’re making them intelligent every step of the way.
When an invoice hits the enterprise, the first question is what’s in it? And today most of the units in shared service environments extract development information via SAP systems. The next question is, “Do I know this supplier?” If they have merged or changed names or opened a new branch, I might not have them in my database. That’s a fuzzy lookup. The next step might be, “Have I ordered something like this?” and that’s a significant question because in some industries up to one-third of spending actually doesn’t have a purchase order. Finding people who have an order of this stuff, all related stuff from this supplier, or similar suppliers in the past, can be the key to figuring out whether we should approve it or not. Then, there’s the question of, “Did we receive the goods and services that this invoice is for?” That’s about going through lists and lists of staff, and figuring out whether the bill of lading for the truck that arrived really contains all the things that were on the truck and all the things that were on the invoice, but no other things. That’s about list matching and list comprehensing, and document matching, and recommending classification systems. It goes on and on like that until the point where we actually put through their payment, and the supplier gets paid for the first invoice that was there.
What you see is a digital process that is enabled by IT systems, very sophisticated IT systems, routine workflows between many human participants today. What you do is we can take the digital exhaust of all the process participants to learn what they’ve been doing, and then put the common, the repetitive, the mind-numbing part of the process on autopilot—gaining speed, reducing cost, making people more satisfied with their work day, because they can focus on the challenging, and the interesting, and the stimulating cases, and increasing customer satisfaction, or in this case supplier satisfaction because they get paid faster. This end-to-end approach is how we look at business processes, and when my ML group and AI do that, we see an order recommender, an entity extractor or some kind of translation mechanism at every step of the process. We work hard to turn these capabilities into scalable APIs on our cloud platform that integrates seamlessly with these standard applications, and that’s really our approach to problem-solving. It ties to the underlying data repository about how business operates and how processes slow.
Did you find that your customers are clear with how this technology can be used, and they’re coming to you and saying, “We want this kind of functionality, and we want to apply it this way,” and they’re very clear about their goals and objectives? Or are you finding that people are still finding their sea legs and figuring out ways to apply artificial intelligence in the business, and you’re more heading to lead them and say, “Here’s a great thing you could do that you maybe didn’t know was possible?”
I think it’s like everywhere, you’ve got early adopters, and innovation promoters, and dealers who actively come with these cases of their own. You have more conservative enterprises looking to see how things play out and what the results for early adopters are. You have others who have legitimate reasons to focus on burning parts of their house right now, for whom this, right now is not yet a priority. What I can say is that the amount of interest in ML and AI that we’re seeing from customers and partners is tremendous and almost unprecedented, because they all see the potential to tag business processes and the way business executes to a complete new level. The key challenge is working with customers early enough, and at the same time working with enough customers in a given setting to make sure that this is not a one-off that is highly specific, and to make sure that we’re really rethinking the process with digital intelligence instead of simply automating the status quo. I think this is maybe the biggest risk. We have tremendous opportunity to transform how business is done today if we truly see this through end-to-end and if we are looking to build out the robots. If we’re only trying to build isolated instances of faster horses, the value won’t be there. This is why we take such an active interest in the end-to-end and integration perspective.
Alright well, I guess just to two final questions. The first is, overall it sounds like you’re optimistic about the transformative power of artificial intelligence and what it can do—
Absolutely Byron.
But I would put that question to you that you put to businesses. What keeps you awake at night? What are the three things that worry you? They don’t have to be big things, but what are the challenges right now that you’re facing or thinking about like, “Oh, I just wish I had better data or if we could just solve this one problem?”
I think the biggest thing keeping me awake right now is the luxury problem of being able to grow as fast as demand and the market wants us to. That has all the aspects of organizational scaling and scaling the product portfolio that we enable with intelligence. Fortunately, we’re not a small start-up with limited resource. We are the leading enterprise software company and scaling inside such an environment is substantially easier than it would be on the outside. Still, we’ve been doubling every year, and we look set to continue in that vein. That’s certainly the biggest strain and the biggest worry that I face. It’s very old-fashioned things; it’s like leadership development that I tend to focus a lot of my time on. I wish I would have more time to play with models, and to play with the technology and to actually build and ship a great product. What keeps me awake is these more old-fashioned things, one of leadership development that matter the most for where we are at right now.
You talked at the very beginning, you said that during the week you’re all about applying these technologies to businesses, and then on the weekend you think about some of these fun problems? I’m curious if you consume science fiction like books or movies, or TV, and if so, is there any view of the future, anything you’ve read or seen or experienced that you think, “Ah, I could see that happening.” Or, “Wow, that really made me think.” Or do you not consume science fiction?
Byron, you caught me out here. The last thing I consumed was actually Valerian and the City of a Thousand Planets just last night in the movie theater in Karlsruhe that I went to all the time when I was a student. While not per se occupied with artificial intelligence, it was certainly stunning, and I do consume a lot of the stuff from the ease of it. It provides a view of plausible futures. Most of the things I tend to read are more focused on things like space, oddly enough. So things like The Three-Body Problem, and the fantastic trilogy that that became, really aroused my interest, and really made me think. There are others that offer very credible trajectories. I was a big fan of the book called Accelerando, which paints a credible trajectory from today’s world of information technology to an upload culture of digital minds and humans colonizing the solar system and beyond. I think that these escapes are critical to cure the hem from day-to-day business, and the pressures of delivering product under a given budget and deadlines. Sort of indulging in them, allows me to return relaxed, and refreshed, and energized on every Monday morning.
Alright, well that’s a great place to leave it, Markus. I’m want to thank you so much for your time. It sounds like you’re doing fantastically interesting work, and I wish you the best.
Did I mention that we’re hiring? There’s a lot of fantastically interesting work here, and we would love to have more people engaging in it. Thank you, Byron.
Byron explores issues around artificial intelligence and conscious computers in his new book The Fourth Age: Smart Robots, Conscious Computers, and the Future of Humanity.
[voices_in_ai_link_back]

Voices in AI – Episode 37: A Conversation with Mike Tamir

[voices_in_ai_byline]
In this episode, Byron and Mike talk about AGI, Turing Test, machine learning, jobs, and Takt.
[podcast_player name=”Episode 37: A Conversation with Mike Tamir” artist=”Byron Reese” album=”Voices in AI” url=”https://voicesinai.s3.amazonaws.com/2018-03-27-(00-55-21)-mike-tamir.mp3″ cover_art_url=”https://voicesinai.com/wp-content/uploads/2018/03/voices-headshot-card-1.jpg”]
[voices_in_ai_byline]
Byron Reese: This is Voices in AI, brought to you by Gigaom. I’m Byron Reese. I’m excited today, our guest is Mike Tamir. He is the Chief Data Science Officer at Takt, and he’s also a lecturer at UC Berkeley. If you look him up online and read what people have to say about him, you notice that some really, really smart people say Mike is the smartest person they know. Which implies one of two things: Either he really is that awesome, or he has dirt on people and is not above using it to get good accolades. Welcome to the show, Mike!
Mark Cuban came to Austin, where we’re based, and gave a talk at South By Southwest where he said the first trillionaires are going to be in artificial intelligence. And he said something very interesting, that if he was going to do it all over again, he’d study philosophy as an undergrad, and then get into artificial intelligence. You studied philosophy at Columbia, is that true?
I did, and also my graduate degree, actually, was a philosophy degree, cross-discipline with mathematical physics.
So how does that work? What was your thinking? Way back in the day, did you know you were going to end up where you were, and this was useful? That’s a pretty fascinating path, so I’m curious, what changed, you know, from 18-year-old Mike to today?
[Laughs] Almost everything. So, yeah, I think I can safely say I had no idea that I was going to be a data scientist when I went to grad school. In fact, I can safely say that the profession of data science didn’t exist when I went to grad school. I did, like a lot of people, who joined the field around when I did, kind of became a data scientist by accident. My degree, while it was philosophy, was fairly technical. It made me more focused on mathematical physics and helped me learn a little bit about machine learning while I was doing that.
Would you say studying philosophy has helped you in your current career at all? I’m curious about that.
Um, well, I hope so. It was very much a focus thing, the philosophy of science. So I think back all the time when we are designing experiments, when we are putting together different tests for different machine learning algorithms. I do think about what is a scientifically-sound way of approaching it. That’s as much the physics background as it is the philosophy background. But it certainly does influence, I’d say daily what we do in our data science work.
Even being a physicist that got into machine learning, how did that come about?
Well, a lot of my graduate research in physics was focused on a little bit of neural activity, but also a good deal of it was focusing in quantum statistical mechanics, which really involved doing simulations and thinking about the world in terms of lots of random variables and unknowns that results in these emergent patterns. And in a lot of ways what we do now, in fact, at Takt is actually writing a lot about group theory and how that can be used as a tool for analyzing the effectiveness of deep learning. Um, there are a lot of, at least at a high level, similarities in trying to find those superpatterns in the signal in machine learning and the way you might think about emergent phenomenon in physical systems.
Would an AGI be emergent? Or is it going to be just nuts and bolts brute force?
[Laughs] That is an important question. The more I find out about successes, at least the partial successes, that can happen with deep learning and with trying to recreate the sorts of sensitivities that humans have, that you would have with object recognition, with speech recognition, with semantics, with general, natural language understanding, the more sobering it is thinking about what humans can do, and what we do with our actual, with our natural intelligence, so to speak.
So do you think it’s emergent?
You know, I’m hesitant to commit. It’s fair to say that there is something like emergence there.
You know this subject, of course, a thousand times better than me, but my understanding of emergence is that there are two kinds: there’s a weak kind and a strong one. A weak one is where something happens that was kind of surprising—like you could study oxygen all your life, and study hydrogen but not be able to realize, “Oh, you put those together and it’s wet.” And then there’s strong emergence which is something that happens that is not deconstructable down to its individual components, it’s something that you can’t actually get to by building up—it’s not reductionist. Do you think strong emergence exists?
Yeah, that’s a very good question and one that I refuse to think about quite a bit. The answer, or my answer I think would be, it’s not as stark as it might seem. Most cases of strong emergence that you might point to, actually, there are stories you can tell where it’s not as much of a category distinction or a non-reducible phenomenon as you might think. And that goes for things as well studied as space transitions, and criticality phenomenon in the physics realm, as it does possibly for what we talk about when we talk about intelligence.
I’ll only ask you one more question on this, and then we’ll launch into AI. Do you have an opinion on whether consciousness is a strong emergent phenomenon? Because that’s going to speak to whether we can build it.
Yeah so, that’s a very good question, again. I think that what we find out when we are able to recreate some of the—we’re really just in the beginning stages in a lot of cases—at least semi-intelligent, or a component of what integrated AI look like. It shows more about the magic that we see when we see consciousness. It brings human consciousness closer to what we see in the machines rather than the other way around.
That is to say, human consciousness is certainly remarkable, and is something that feels very special and very different from what maybe imperatively constructed machine instructions are. There is another way of looking at it though, which is that maybe by seeing how, say, a deep neural net is able to adapt to signals that are very sophisticated and maybe even almost impossible to really boil it down, it’s actually something that we do that we might imagine are brains are doing all the time, just in a far, far larger magnitude of parameters and network connections.
So, it sounds like you’re saying it may not be that machines are somehow ennobled with consciousness, but that we discover that we’re not actually conscious. Is that kind of what you’re saying?
Yeah, or maybe something in the middle.
Okay.
Certainly, our personal experience of consciousness, and what we see when we interact with other humans or other people, more generally; there’s no denying that, and I don’t want to discount how special that is. At the same time, I think that there is a much blurrier line, is the best way to put it, between artificial, or at least the artificial intelligence that we are just now starting to get our arms around, and what we actually see naturally.
So, the shows called Voices in AI, so I guess I need to get over there to that topic. Let’s start with a really simple question: What is artificial intelligence?
Hmm. So, until a couple years ago, I would say that artificial intelligence really is what we maybe now call integrated AI. So a dream of using maybe several integrated techniques of machine learning to create something that we might mistake for, or even accurately describe as, consciousness.
Nowadays, the term “artificial intelligence” has, I’d say, probably been a little bit whitewashed or diluted. You know, artificial intelligence can mean any sort of machine learning or maybe even no machine learning at all. It’s a term that a lot of companies put in their VC deck, and it could be something as simple as just using a logistic regression—hopefully, logistic regression that uses gradient descendants as opposed to closed-form solution. Right now, I think it’s become kind of indistinguishable from generic machine learning.
I, obviously, agree, but, take just the idea that you have in your head that you think is legit: is it artificial in the sense that artificial turf isn’t really grass, it just looks like it? Or is it artificial in the sense we made it. In other words, is it really intelligence, or is it just something that looks like intelligence?
Yeah, I’m sure people bring up the Turing test quite a bit when you broach this subject. You know, the Turing test is very coarsely… You know, how would you even know? How would you know the difference between something that is an artificial intelligence and something that’s a bona fide intelligence, whatever bona fide means. I think Turing’s point, or one way of thinking about Turing’s point, is that there’s really no way of telling what natural intelligence is.
And that again makes my point, that it’s a very blurry line, the difference between true or magic soul-derived consciousness, and what can be constructed maybe with machines, there’s not a bright distinction there. And I think maybe what’s really important is that we probably shouldn’t discount ostensible intelligence that can happen with machines, any more than we should discount intelligence that we observe in humans.
Yeah, Turing actually said, a machine may do it differently but we still have to say that the machine is thinking, it just may be different. He, I think, would definitely say it’s really smart, it’s really intelligent. Now of course the problem is we don’t have a consensus definition even of intelligence, so, it’s almost intractable.
If somebody asks you what’s the state of the art right now, where are we at? Henceforth, we’re just going to use your idea of what actual artificial intelligence is. So, if somebody said “Where are we at?” are we just starting, or are we actually doing some pretty incredible things, and we’re on our way to doing even more incredible things?
[Laughs] My answer is, both. We are just starting. That being said, we are far, we are much, much further along than I would have guessed.
When do you date, kind of, the end of the winter? Was there a watershed event or a technique? Or was it a gradualism based on, “Hey, we got faster processors, better algorithms, more data”? Like, was there a moment when the world shifted? 
Maybe harkening back to the discussion earlier, you know, someone who comes from physics, there’s what we call the “miracle year,” when Einstein published his theory—a really remarkable paper—roughly just over a hundred years ago. You know, there is a miracle year and then there’s also when he finally was able to crack the code in general relativity. I don’t think we can safely say that there been a miracle year until far, far in the future, when it comes to the realm of deep learning and artificial intelligence.
I can say that, in particular, with natural language understanding, the ability to create machines that can capture semantics, the ability of machines to identify objects and to identify sounds and turn them into words, that’s important. The ability for us to create algorithms that are able to solve difficult tasks, that’s also important. But probably at the core of it is the ability for us to train machines to understand concepts, to understand language, and to assign semantics effectively. One of the big pushes that’s happened, I think, in the last several years, when it comes to that, is the ability to represent sequences of terms and sentences and entire paragraphs, in a rich mathematically-representable way that we can then do things with. That’s been a big leap, and we’re seeing a lot of the progress that with neural word embeddings with sentence embeddings. Even as recently as a couple months ago, some of the work with sentence embedding that’s coming out is certainly part of that watershed, and that move from dark ages in trying to represent natural language in a intelligible way, to where we are now. And I think that we’ve only just begun.
There’s been a centuries-old dream in science to represent ideas and words and concepts essentially mathematically, so that they can manipulated just like anything else can be. Is that possible, do you think?
Yeah. So one way of looking at the entire twentieth century is a gross failure in the ability to accurately capture the way humans reason in Boolean logic, and the way we represent first order logic, or more directly in code. That was a failure, and it wasn’t until we started thinking about the way we represent language in terms of the way concepts are actually found in relation to one another, training an algorithm to read all of Wikipedia and to start embedding that with Word2vec—that’s been a big deal.
The fact that by doing that, and now we can start capturing everything. It’s sobering, but we now have algorithms that can, with embed sentences, detect things like logical implications or logical equivalence, or logical non-equivalence. That’s a huge step, and that’s a step that I think we tried quite a bit to do, or many tried to do without experience and failed.
Do you believe that we are on a path to creating an AGI, in the sense that what we need is some advances in algorithms, some faster machines, and more data, and eventually we’re going to get there? Or, is AGI going to come about, if it does, from a presently-unknown approach, a completely different way of thinking about knowledge?
That’s difficult to speculate. Let’s take a step back. Five years ago, less than five years ago, if you wanted to propose a deep learning algorithm for an industry to solve a very practical problem, the response you would get is stop being too academic, let’s focus on something a little simpler, a little bit easier to understand. There’s been a dramatic shift, just in the last couple years, that now, the expectation is if you’re someone in the role that I’m in, or that my colleagues are in, if you’re not considering things like deep learning, then you’re not doing your job. That’s something that seems to have happened overnight, but was really a gradual shift over the past several years.
Does that mean that deep learning is the way? I don’t know. What do you really need in order to create an artificial intelligence? Well, we have a lot of the pieces. You need to be able to observe maybe visually or with sounds. You need to be able to turn those observations into concepts, so you need to be able to do object recognition visually. Deep learning has been very successful in solving those sorts of problems, and doing object recognition, and more recently making that object recognition more stable under adversarial perturbation.
You need to be able to possibly hear and respond, and that’s something that we’ve gotten a lot better at, too. We’ve got a lot of the work done by doing research labs, there’s been some fantastic work in making that more effective. You need to be able to not just identify those words or those concepts, but also put them together, and put them together, not just in isolation but in the context of sentences. So, the work that’s coming out of Stanford and some of the Stanford graduates, Einstein Labs, which is sort of at the forefront there, is doing a very good job in capturing not just semantics—in the sense of, what is represented in this paragraph and how can I pull out the most important terms?—but doing a job of abstractive text summarization, and, you know, being able to boil it down to terms and concepts that weren’t even in the paragraph. And you need to be able to do some sort of reasoning. Just like the example I gave before, you need to be able to use sentence embedding to be able to classify—we’re not there yet, but—that this sentence is related to this sentence, and this sentence might even entail this sentence.
And, of course, if you want to create Cylons, so to speak, you also need to be able to do physical interactions. All of these solutions in many ways have to do with the general genre of what’s now called “deep learning,” of being able to add parameters upon parameters upon parameters to your algorithm, so that you can really capture what’s going on in these very sophisticated, very high dimensional spaces of tasks to solve.
No one’s really gotten to the point where they can integrate all of these together, and I think is that going to be something that is now very generic, that we call deep learning, which is really a host of lots of different techniques that just use high dimensional parameter spaces, or is it going to be something completely new? I wouldn’t be able to guess.
So, there are a few things you left of your list, though, so presumably you don’t think an AGI would need to be conscious. Consciousness isn’t a part of our general intelligence. 
Ah, well, you know, maybe that brings us back to where we started.
Right, right. Well how about creativity? That wasn’t in your list either. Is that just computational from those basic elements you were talking about? Seeing, recognizing, combining?
So, an important part of that is being able to work with language, I’d say, being able to do natural language understanding and do natural language understanding at higher than the word level, but at the sentence level, certainly anything that might be what they call mistaken or “identified as” thinking. Have to have that as a necessary component. And being able to interact, being able to hold conversations, to abstract and to draw conclusions and inferences that aren’t necessarily there.
I’d say that that’s probably the sort of thing that you would expect of a conscious intelligence, whether it’s manifest in a person or manifest in a machine. Maybe I should say manifested in a human, or manifested in a machine.
So, you mentioned the Turing test earlier. And, you know, there are a lot of people who build chatbots and things that, you know, are not there yet, but people are working on it. And I always type in one, first question, it’s always the same, and I’ve never seen a system that even gets the question, let alone can answer it.
The question is, “What’s bigger, a nickel or the sun?” So, two questions, one, why is that so hard for a computer, and, two, how will we solve that problem?
Hmm. I can imagine how would I build a chatbot, and I have worked on this sort of project in the past. One of the things—and I mentioned earlier, this allusion to a miracle year—is the advances that happened, in particular, in 2013 with figuring out ways of doing neural-word embeddings. That’s so important, and one way of looking at why that’s so important is that, when we’re doing machine learning in general—this is what I tell my students, this what drives a lot of our design—you have to manage the shape of your data. You have to make sure that the amount of examples you have, the density of data points you have, is commensurate with the amount of degrees of freedom that you have representing your world, your model.
Until very recently, there have been attempts, but none of them as successful as we’ve seen in the last five years. The baseline has been what’s called the one-hot vector encoding, where you have a different dimension for every word in your language, usually it’s around a million words. You have all zeros and then for the word maybe in the first dimension you take the first word in the dictionary to order them that way, and you have the word ‘a,’ which is spelled with the letter ‘a,’ and that’s then the one and all zeros. And then for the second word you have a zero and a one and the rest zeros. So the point here, and not to get technical, but your dimensions are just too many.
You have millions and millions of dimensions. When we talk with students about this, it’s called the curse of dimensionality, every time you add even one dimension, you need twice as many data points in order to maintain the same density. And maintaining that density is what you need in order to abstract, in order to generalize, in order to come up with an algorithm that can actually find a pattern that works, not just for the data that it sees, but for the data that it will see.
What happens with these neural word embeddings? Well, they solve the problem of the curse of dimensionality, or at least they’ve really gotten their arms a lot further around it than ever before. They’ve enabled us to represent terms, represent concepts, not in these million dimensional vector spaces, where all that rich information is still there, but it’s spread so thinly across so many dimensions that you can’t really find a single entity as easily as you can if it were only representing a smaller number of dimensions, and that’s what these embeddings do.
Now, once you have that dimensionality, once you’re able to compress them into a lower dimension, now you can do all sorts of things that you want to do with language that you just couldn’t do before. And that’s part of why we see this slow operation with chatbots, they probably have something like this technology. What does this have to do with your question? These embeddings, for the most part, happen not by getting instructions—well nickels are this size, and they’re round, and they’re made of this sort of composite, and they have a picture of Jefferson stamped on the top—that’s not how you learn to mathematically represent these words at all.
What you do is you feed the algorithm lots and lots of examples of usage—you let it read all of Wikipedia, you let it read all of Reuters—and slowly but surely what happens is, the algorithm will start to see these patterns of co-usage, and will start to learn how one word follows after another. And what’s really remarkable, and could be profound, at least I know that a lot of people would want to infer that, is that the semantic kind of comes out for free.
You end up seeing the geometry of the way these words are embedded in such a way that you see, a famous example is a king vector minus a man vector plus a woman vector equals a queen vector, and that actually bears out in how the machine can now represent the language, and it did that without knowing anything about men, women, kings, or queens. It did it just by looking at frequencies of occurrence, how those words occur next to each other. So, when you talk about nickels and the sun, my first thought, given that running start, is that well, the machine probably hasn’t seen a nickel and a sun in context too frequently, and one of the dirty secrets about these neural embeddings is that they don’t do as well on very low-frequency terms, and they don’t always do well in being able to embed low frequency co-occurrences.
And maybe it’s just the fact that it hasn’t really learnt about, so to speak, it hasn’t read about, nickels and suns in context together.
So, is it an added wrinkle that, for example, you take a word like set, s-e-t, I think OED has two or three hundred definitions of it, you know—it’s something you do, it’s an object, etcetera. You know there’s a Wikipedia entry on a sentence, an eight word long grammatically correct sentence which is, “Buffalo buffalo buffalo buffalo buffalo buffalo buffalo buffalo,” which contains nouns, verbs, all of that. Is there any hope that if you took all the monkeys in all the universe typing cogent and coherent sentences, would it ever be enough to train it to what a human can do?
There’s a couple things there, and one of the key points that you’re making is that there are homonyms in our language, and so work should be done on disambiguating the homonyms. And it’s a serious problem for any natural language understanding project. And, you know, there are some examples out there of that. There’s one recently which is aimed at not just identifying a word but also disambiguating the usages or the context.
There are also others, not just focused on how to mathematically-represent how to pinpoint a representation of a word, but also how to represent the breadth of the usage. So maybe imagine not a vector, but a distribution or a cloud, that’s maybe a little thicker as a focal point, and all of those I think are a step in the right direction for capturing what is probably more representative of how we use language. And disambiguation, in particular with homonyms, is a part of that.
I only have a couple more questions in this highly theoretical realm, then I want to get down to the nitty gritty. I’m not going to ask you to pick dates or anything, but the nickel and the sun example, if you were just going to throw a number out, how many years is it until I type that question in something, and it answers it? Is that like, oh yeah we could do it if we wanted to, it’s just not a big deal, maybe give it a year? Or, is it like, “Oh, no that’s kind of tricky, wait five years probably.”
I think I remember hearing once never make a prediction.
Right, right. Well, just, is that a hard problem to solve?
The nickel and the sun is something that I’d hesitate to say is solvable in my lifetime, just to give a benchmark there, violating that maxim. I can’t say exactly when, what I can say is that the speed with which we are solving problems that I thought would take a lot longer to solve, is accelerating.
To me, while it’s a difficult problem and there are several challenges, we are still just scratching the surface in natural language understanding and word representation in particular, you know words-in-context representation. I am optimistic.
So, final question in this realm, I’m going to ask you my hard Turing test question, I wouldn’t even give this to a bot. And this one doesn’t play with language at all.
Dr. Smith is eating lunch at his favorite restaurant. He receives a call, takes it and runs out without paying his tab. Is management likely to prosecute? So you have to be able to infer it’s his favorite restaurant, they probably know who he is, he’s a doctor, that call was probably an emergency call. No, they’re not going to prosecute because that’s, you know, an understandable thing. Like, that doesn’t have any words that are ambiguous, and yet it’s an incredibly hard problem, isn’t it?
It is, and in fact, I think that is the, that is one of the true benchmarks—even moreso than comparing a nickel and a sun—of real, genuine natural language understanding. It has all sorts of things—it has object permanence, it has tracking those objects throughout different sentences, it has orienting sequences of events, it has management, which is mentioned in that last sentence, which is how you would be able to infer that management is somehow connected to the management of the restaurant.
That is a super hard one to solve for any Turing machine. It’s also something we’re starting to make progress on. Using LSDMs that do several passes through a sequence of sentences, classic artificial sentence dataset, that natural language understanding finds—the Facebook of AGI dataset, which actually is out there to help use as a benchmark for training these sorts of object permanence in multi-sentence thread. And we’ve made modest gains in that. There are algorithms like the Ask Me Anything algorithm, that have shown that it’s at least possible to start tracking objects over time, and with several passes come up with the right answer to questions about objects in sentences across several different statements.
Pulling back to the here and now, and what’s possible and what’s not. Did you ever expect AI to become part of the daily conversation, just to be part of popular culture the way it is now?
About as much as I expect that in a couple years that AI is going to be a term much like Big Data, which is to say overused.
Right.
I think, with respect to an earlier comments, the sort of AI that you and I have been dancing around, which is fully-integrated AI, is not what we talk about when we talk about what’s in daily conversation now, or for the most part not what we’re talking about in this context. And so it might be a little bit of a false success, or a spurious usage of “AI” in as much frequency as we see it.
That doesn’t mean that we haven’t made remarkable advances. It doesn’t mean that the examples that I’ve mentioned, in particular, in deep learning aren’t important, and aren’t very plausibly an early set of steps on the path. I do think that it’s a little bit of hype, though.
If you were a business person and you’re hearing all of this talk, and you want to do something that’s real, and that’s actionable, and you walk around your business, department to department—you go to HR, and to Marketing and you got to Sales, and Development—how do you spot something that would be a good candidate for the tools we have today, something that is real and actionable and not hype?
Ah, well, I feel like that is the job I do all the time. We’re constantly meeting with new companies, Fortune 500 CEOs and C-Suite execs, and talking about the problems that they want to solve, and thinking about ways of solving them. Like, I think a best practice is to always to keep it simple. There are a host of free deep learning techniques for doing all sorts of things—classification, clustering, user item matching—that are still tried-and-true, and that should probably be done first.
And then there are now, a lot of great paths to using these more sophisticated algorithms that mean that you should be considering them early. How exactly to consider one case from the other, I think that part of that is practice. It’s actually one of the things that when I talk to students about what they’re learning, I find that they’re walking away with not just, “I know what the algorithm is, I know what the objective function is, and how to manage momentum in the right way and optimizing that function,” but also how do you see the similarity between matching users and items in the recommender, or abstracting the latent semantic association of a bit of text or with an image, and there are similarities, and certain algorithms that solve all those problems. And that’s, in a lot of ways, practice.
You know, when the consumer web first came out and it became popularized, people had, you know, a web department, which would be a crazy thought today, right? Everything I’ve read about you, everybody says that you’re practical. So, from a practical standpoint, do you think that companies ought to have an AI taskforce? And have somebody whose job it is to do that? Or, is it more the kind of thing that it’s going to gradually come department by department by department? Or, is it prudent to put all of your thinking in one war room, as it were?
So, yeah, the general question is what’s the best way to do organizational design with machine learning machines, and the first answer is there are several right ways and there are a couple wrong ways. So, one of these wrong ways of the early-days are where you have this data science team that is completely isolated and is only responsible for R&D work, prototyping certain use cases and then they, to use a phrase you hear often, throw it over the wall to engineering to go implement, because I’m done with this project. That’s a wrong way.
There are several right ways, and those right ways usually involve bringing the people who are working on machine learning closer to production, closer to engineering, and also bringing the people involved in engineering and production closer to the machine learning. So, overall blurring those lines. You can do this with vertical integrated small teams, you could do this with peer teams, you can do this with a mandate that some larger companies, like Google, are really focused on making all their engineers machine learning engineers. I think all those strategies can work.
It all sort of depends on the size and the context of your business, and what kind of issues you have. And depending on those variables, then, among the several solutions, there might be one or two that are most optimal.
You’re the Chief Data Science Officer at Takt, spelled T-A-K-T, and is takt.com if anybody wants to go there. What does Takt do?
So we do the backend machine learning for large-scale enterprises. So, you know, many of your listeners might go to Starbucks and use the app to pay for Starbucks coffee. We do all of the machine learning personalization for the offers, for the games, for the recommendors in that app. And the way we approach that is by creating a whole host of different algorithms for different use cases—this goes back to your earlier question of abstracting those same techniques for many different use cases—and then apply that for each individual customer. We find the list completion use case, the recursive neural network approach, where there’s a time series of opportunity, where you can have interactions with an end user, and then learn from that interaction, and follow up with another interaction, doing things like reinforcement learning to do several interactions in a row, which may or may not get a signal back, but we have been trained to work towards that goal over time without that direct feedback signal.
This is the same sort of algorithms, for instance, that were used to train AlphaGo, to win a game. You only get that feedback at the end of the game, when you’ve won or lost. We take all of those different techniques and embed them in different ways for these large enterprise customers.
Are you a product company, a service company, a SaaS company—how does all that manifest?
We are a product company. We do tend to focus on the larger enterprises, which means that there is a little bit of customization involved, but there’s always going to be some customization involved when it comes to machine learning. Unless it’s just a suite of tools, which we are not. And what that means is that you do have to train and apply and suggest the right kinds of use cases for the suite of tools that we have, machine learning tools that we have.
Two more questions, if I may. You mentioned Cylons earlier, a Battlestar Galactica reference to those who don’t necessarily watch it. What science fiction do you think gets the future right? Like, when you watch it or read it, or what have you, you think “Oh yeah, things could happen that way, I see that”?
[Laughs] Well, you know the physicist in me still is both hopeful and skeptical about faster-than-light travel, so I suppose that wouldn’t really be the point of your question, is more with computers and with artificial intelligence.
Right, like Her or Ex Machina or what have you.
You know, it’s tough to say which of these, like, conscious-being robots is the most accurate. I think there are scenes worth observing that already have happened. Star Trek, you know, we create the iPad way before they had them in Star Trek time, so, good for reality. We also have all sorts of devices. I remember, when, in the ’80s—to date myself—the movie Star Trek came out, and Scotty gets up in front of his computer, an ’80s computer, and picks up the mouse and starts speaking into it and saying, “Computer, please do this.”
And my son will not get that joke, because he can say “Hey, Siri” or “Okay, Google” or “Alexa” or whatever the device is, and the computer will respond. And that’s, I like to focus on those smaller wins, that we are dramatically much quicker than forecasts in some cases able to accomplish that. I did see an example the other day about HAL, the Space Odyssey artificial intelligence, where people were mystified that this computer program could beat a human in chess, but didn’t blink an eye that the computer program could not only hold a conversation, but has a very sardonic disposition towards the main character. That, probably, very well captures this dichotomy of the several things are very likely to be captured, and we can get to very quickly, and other things that we thought were easy but take quite a lot longer than expected.
Final question, overall, are you an optimist? People worry about this technology—not just the killer robots scenario, but they worry about jobs and whatnot—but what do you think? Broadly speaking, as this technology unfolds, do you see us going down a dystopian path, or are you optimistic about the future?
I’ve spoken about this before a little bit. I don’t want to say, “I hope,” but I hope that Skynet will not launch a bunch of nuclear missiles. I can’t really speak with confidence to whether that’s a true risk or just an exciting storyline. What I can say is that the displacement of service jobs by automated machines is a very clear and imminent reality.
And that’s something that I’d like to think that politicians and governments and everybody should be thinking about—in particular how we think about education. The most important skill we can give our children is teaching them how to code, how to understand how computer programs work, and that’s something that we really just are not doing enough of yet.
And so will Skynet nuke everybody? I don’t know. Is it the case that I am, at six years old, teaching my son how to code already? Absolutely. And I think that will be make a big difference in the future.
But wouldn’t coding be something relatively easy for an AI? I mean it’s just natural language, tell it what you want it to do.
Computers that program themselves. It’s a good question.
So you’re not going to suggest, I think you mentioned, your son be a philosophy major at Columbia?
[Laughs] You know what, as long as he knows some math and he knows how to code, he can do whatever he wants.
Alright, well we’ll leave it on that note, this was absolutely fascinating, Mike. I want to thank you, thank you so much for taking the time. 
Well thank you, this was fun.
Byron explores issues around artificial intelligence and conscious computers in his upcoming book The Fourth Age, to be published in April by Atria, an imprint of Simon & Schuster. Pre-order a copy here.
[voices_in_ai_link_back]

Recent Enterprise File Sync and Sharing News

Here is a brief round-up of some recent news from the Enterprise File Synchronization and Sharing market segment.

EFSS Application Security

MobileIron published a whitepaper, titled “State of App Security”, that includes results of a survey conducted with its customers. The survey and white paper are briefly summarized in this post.
Survey respondents were asked to list the cloud applications that had been blacklisted by their IT departments. Of the top ten apps listed, five were EFSS solutions: Dropbox, Microsoft OneDrive, Google Drive, Box, and SugarSync.
It’s important to note that all of these blacklisted apps are consumer-oriented and their vendors do offer business versions that are not commonly blacklisted because they include better security features. However, the unauthorized or “shadow” use of consumer EFSS solutions within businesses continues to pose significant information security risks.

Dropbox Doubles Down on Business

Dropbox made several product and business strategy announcements at its inaugural customer event, Dropbox Open, which was held on November 4th, in San Francisco. Most were directly relevant to the company’s increasing focus on businesses, rather than consumers. They are  briefly summarized in this Dropbox post, but here’s the skinny on a few.
First, it’s clear why Dropbox is doubling down on its efforts to win over organizations. The company announced that it has signed up around 50,000 new organizations as paying Dropbox Business customers in the last year. Dropbox now claims to have 150,000 business customers; that’s organizations, not seats. The company stated that business is it’s fastest growing target market.
To underscore the point, Dropbox announced a new product, Dropbox Enterprise, which “provides the same core security features, admin capabilities, and modern collaboration tools as Dropbox Business — plus new deployment tools, advanced controls, and services and support designed specifically for large organizations.”
Dropbox also announced three new administrative features that will be included in Dropbox Business as well as in Dropbox Enterprise. The new capabilities ‒ suspended user state, sign in as user, and custom branding ‒ are available now through the company’s Early Access program, with no general release date given.
Dropbox is going down the same road that Box has already traveled. It started with a consumer grade product, added functionality to make it more attractive and useful for small and medium businesses, and now is incorporating the robust security and control features that IT departments in large enterprises demand. The big question now is can Dropbox overtake Box in the EFSS market?

Google Drive Adds New Features

Google announced three new capabilities that are intended to improve the usability of Google Drive. These new features apply to all Google Drive users, not just business employees.
It’s now possible to receive a notification from the application on your Android or iOS device when someone has shared a file or folder with you. Previously, those notifications were made via email. The new notifications are actionable; clicking the link will take you to the document or folder that has bee shared.
Google Drive users can now request and grant access to a file or folder to which a link has been sent, but the owner forgot to extend access rights. The feature is mobile friendly. Android users can request access with a single tap. File and folder owners can instantly be notified of the request and provide access from their Android or iOS device.
Finally, it’s now possible to preview files stored on Google Drive on Android devices even if you don’t have a Google account. That feature has been available in Web browsers for a while and makes sense in that context. It’s hard to imagine why an Android device owner wouldn’t have a Google account, but, apparently, its is a problem and Google chose to address it.

Syncplicity Plays Catch-Up on Mobile Security

Syncplicity announced partnerships with AirWatch and MobileIron to help customers secure files on mobile devices. It should be safe to assume that the integration with AirWatch had been ready (or nearly so) for quite a while, since both were owned by EMC until it spun off Syncplicity a couple of months ago. At any rate, these partnerships merely bring Syncplicity even with its competitors, who have had similar partnerships or their own mobile device containerization capabilities for some time now.

Box Expands Its European Presence

Box has opened two new offices in Europe in the last 3 weeks, one in Amsterdam and another in Stockholm. This continental presence is crucial to Box as it seeks to grow by expanding overseas sales efforts. However, the new offices also raise questions about how Box (and competitors) will deal with the recent nullification of the Safe Harbor agreement that had been in place between the European Union and the United States.

ownCloud Brings Control of Open Source EFSS On-Premises

ownCloud announced the newest version (8.2) of its open source EFFS offering, which moves it to a hybrid model. With ownCloud 8.2, it’s now possible for customers to deliver security and control of their files residing in the cloud through an on-premises adminstrative console.

Linoma GoDrive Customers Gain Mobile Access

In another transformation to a vendor’s existing EFSS model, Linoma Software unveiled its GoAnywhere mobile apps for its GoDrive on-premises EFSS solution. Linoma customers can now access files residing in GoDrive from iOS and Android mobile devices. While files and folder are encrypted during transit, Linoma does not secure files while they are on a mobile device. However, they do provide an administrative capability to deactivate and wipe files and folders from devices that have been lost or stolen.

This Digital Transformation is Not the One You’re Looking For

I was sorting through some browser tabs that had been open for a couple of weeks on my laptop and rediscovered a press release that had caught my attention earlier. After rereading it, I realized that I had left the release up in my browser because it could be the poster child for the inane manner in which technology vendors and IT consulting firms are talking about and selling what they very much want to be the next big thing – Digital Transformation.
CA Technologies’ press release was a horrific example right from the start. It’s title, “CA Technologies Study Reveals Widespread Adoption of Digital Transformation”, nearly made me spit coffee all over my laptop. Really? Is Digital Transformation (DT) something that can be adopted? Hardly. After all, DT is not a discrete technology. Rather, it’s a never-ending journey that organizations undertake to better the efficiency and effectiveness of their operations.
DT involves making changes to business objectives, strategies, models, cultures, processes and so many other elements. Many of those changes can be supported by the deployment and adoption of enabling technologies, but DT isn’t about the technology itself. It’s a mindset, a way of thinking and acting as an organization that spans across all of its planning and execution.
In that regard, DT is very much like the discipline known as Knowledge Management (KM) that was similarly a darling of technology vendors and their consulting partners nearly 20 years ago. Most large enterprises at least considered implementing KM practices and technologies. In fact, many did, although the majority of those ‘efforts’ failed to survive an initial pilot program. In the end, only a few big companies, the ones that treated KM as something more than a technology set to be adopted, whole-heartedly embraced the discipline and successfully wove it into nearly every aspect of their businesses.
We’ve seen the same phenomenon play out with Social Business. McKinsey & Company has been tracking the deployment and impact of social constructs, behaviors and tools in a cohort of roughly 1,500 enterprises for nearly 10 years now. Earlier this month, in a teaser to its complete report of annual survey results, McKinsey published these related and telling findings:

“…35 percent of the companies had adopted social technologies in response to their adoption by competitors. Copycat behavior was also responsible for their diffusion within organizations, though at a slightly lower rate: 25 percent of all employee usage. Roughly a fifth of the companies we studied will account for an estimated 50 percent of all social-technology usage in 2015.”

Most organizations and individuals tried to ‘adopt’ social technologies because they felt competitive pressure to do so (thanks, in part, to vendors and consultants), not because they had investigated and understood how ‘being social’ at work could change how well their organization actually performed relative to both its current state and its competitors. On the other hand, a minority of organizations (20% in McKinsey’s survey) have made the dedicated, all-in commitment needed to succeed with Social Business.
Today, we are beginning this cycle all over again, this time under the moniker of Digital Transformation. Consider these findings from CA’s study:

“Digital Transformation is being driven as a coordinated strategy across a majority of organizations (55 percent)…  As a result, 45 percent of respondents have already seen measurable increases in customer retention and acquisition from their digital transformation initiatives and 44 percent have seen an overall increase in revenue.”

In other words, if you aren’t “adopting” DT already, you’re toast. At least that’s what CA and other technology vendors and consultants want you to believe in a fresh state of panic. Hence these findings from CA’s study:

Digital Disrupters have two times higher revenue growth than mainstream organizations. They report two and a half times higher profit growth than the mainstream organizations.”

That may be accurate, but surely those “Digital Disrupters” did not achieve the reported results merely by adopting technology, whether it be from CA or another vendor. They’re the ones who have taken a comprehensive view of DT and, as CA itself puts it, have “…many projects underway in multiple areas of the company, including customer services, sales and marketing, and product/service development.” It’s not a coincidence that CA was only able to include 14% of the organizations surveyed in the group it labeled “Digital Disrupters”. That matches up pretty well with McKinsey’s finding of just 20% of organizations surveyed making more than a token effort at becoming a social business.
All of this is to say beware of vendors and consultants selling technology as the cornerstone of DT initiatives. Yes, technology is an invaluable piece of the puzzle, but it’s not the only or most important one. DT can’t simply be adopted; every aspect of it must be considered and actively embraced by the entire organization.

The Internet of Things and Networks of Everything

The Internet of Things (IoT) has been a hot topic for several months now, and there are new stories about it in the business and technology press on a daily basis. While it’s easy to view these as hype at worst and vision at best, there is no denying that purveyors of hardware, software and services are dedicating and creating the resources they will use to capitalize on the IoT. Last week alone, there were three announcements that show just how quickly the IoT market is progressing and how big of a business opportunity it is.
On Monday, September 14th, IBM formally launched a distinct IoT business unit and named former Thomas Cook Group CEO Harriet Green as its leader. The new IoT unit is the first significant step by IBM toward delivering on the $3 billion commitment it made to IoT in March. IBM signaled in Monday’s press release that the unit will “soon” number about 2,000 consultants, researchers and developers, who will use IBM’s assets to help customers get up and running on the IoT. Those assets will likely include the Bluemix platform-as-a-service (PaaS), Watson and other analytics software, as well as the MQTT messaging protocol standard for machine-to-machine communication that IBM submitted to OASIS in 2013.
The next day, Salesforce.com used its annual Dreamforce conference as the grand stage on which to unveil its IoT Cloud. This offering has at its core a new “massively scalable”, real-time event processing engine named ‘Thunder’ (to complement Salesforce’s ‘Lightening’ UI framework). IoT Cloud connects IoT resources and Thunder rules-based workflow to route data between them, triggering pre-defined actions. For example, when an individual enters a retail store, a beacon can offer them discounts based on qualification criterion such as loyalty program status and in-store inventory levels. Scenarios such as this will be possible because of IoT Cloud’s integration with the Salesforce Sales, Marketing and Analytics Clouds. IoT Cloud is currently in pilot and is expected to be generally available sometime in the second half of 2016.
While these two announcements are important milestones in the respective organizations ability to help customers connect to and use the IoT, they do not enable them to do so immediately and risk being labeled as more IoT hype. The sheer magnitude of resources assembled for each of these vendors initiatives signals that they believe that the IoT will be both real and profitable in the not-so-distant future.
The final piece of related news from last week underscores that smaller, pure-play vendors are delivering tools that help their customers get on the IoT now. Build.io announced that Flow, its integration PaaS that had been beta released in March, is now generally available. Flow features a drag-and-drop interface that is used to connect IoT elements ─ sensors and other intelligent devices, backend systems, mobile applications and other software ─ into an integrated system. Connections are made at the API level. Like Salesforce’s Thunder, Flow uses rules-based event processing to trigger actions from IoT data. In essence, Build.io is delivering today a critical part of what Salesforce intends to make generally available later this year.

Current State of the Internet of Things and Networks of Everything

These announcements, taken together, mean that the IoT is poised for takeoff. The first sets of user-friendly tools that organizations need to connect IoT nodes, transmit their data and use it to drive business processes are available now, in some cases, or will be coming to market within a year. We are on the cusp of a rapid acceleration in the growth of the market for software underpinning the IoT, as well as the network itself.
This latest batch of IoT announcements from software vendors underscores another thing: the IoT will initially be built separately from enterprise social networks (ESNs). Many organizations, particularly large enterprises, have experimented with ESNs and a few have managed to build ones that are operating at scale and creating value. Those businesses will be turning their attention to IoT development now, if they haven’t already. They will pilot, then scale, their efforts there, just as they did with ESNs.
Eventually, organizations will realize that it is more efficient and effective to build Networks of Everything (NoE), in which humans and machines communicate and collaborate with one another using not only the Internet, but also cellular, Bluetooth, NFC, RFID and other types of networks. This construct is just beginning to enter reality, and it will take a few years before NoE get the market attention that ESNs did five years ago and the IoT is now.
At some future point, when NoE have become a fixture of networked business, we will look back at this month (Sept. 2015) and declare that it was a watershed moment in the development of the IoT. We’ll also laugh at how obvious it seems, in hindsight, that we should have just built NoE in the first place.

The State of Salesforce Community Cloud

It’s the eve of Salesforce.com’s annual Dreamforce event, and I have the company and its customers on my mind. I’ll be attending Dreamforce again (Disclaimer: Salesforce is covering my registration, travel and hotel expenses). As always, I’ll be taking in all the announcements at Dreamforce, but paying the most attention to the Community Cloud, individual applications and platform components that make up the Salesforce collaboration and content management ecosystem.
Before Dreamforce begins, it’s useful to think about the actual state of collaboration amongst Salesforce’s customer base. There will be marquee customers on stage this week talking enthusiastically about their cutting-edge use of Salesforce’s latest offering versions, including those that are not yet generally available. But what about the mainstream Salesforce customer and how they’re using the company’s products to collaborate?
To get a sense of that, I digested The State of Salesforce survey report that was recently published by Bluewolf, a global consultancy that designs customer-facing, digital experiences using third-party, cloud-based software. This year’s report is the 4th annual edition published by Bluewolf, who surveyed more that 1,500 Salesforce customer organizations, of varied organizational size and located around the world.
Bluewolf’s report does not investigate every bit of Salesforce’s collaboration and content management functionality in detail. Instead, it focuses on the assembled collection of those that is the Community Cloud. In two pages of The State of Salesforce, Bluewolf reports on Salesforce customers’ adoption of Community Cloud, its most common use cases and the high-level business benefits that customers attribute to its use.

Community Cloud Adoption

Of the Salesforce customer companies that have purchased Service Cloud, Sales Cloud, and Marketing Cloud, 36% have also purchased Community Cloud. That represents decent adoption by Salesforce’s best customers, especially for an offering that has only been in-market for a year. Even better, 21% of respondents that already license those other Salesforce clouds said that they plan on purchasing Community Cloud in the coming year. If that pans out, then over half of Salesforce’s most dedicated customers will be on Community Cloud within two years of its launch.
What the report doesn’t illuminate, and I’ll try to investigate at Dreamforce this week, is Community Cloud adoption by the rest of the existing Salesforce customer base. It’s likely that the bar is set much lower there and that Salesforce will need to refocus its marketing and sales of Community Cloud for the next wave of potential adopters. Selling Community Cloud as an enhancement of the other Salesforce clouds is very different than convincing organizations of its utility as an independent collaboration and content management solution.

Community Cloud Use Cases

As for Community Cloud use cases, Bluewolf’s survey found that the top three were Customer Service (25% of respondents), Partner Enablement (21%) and Internal Collaboration (17%). Given Salesforce’s current positioning as “The Customer Success Platform”, and the amount of resources it has spent to launch and grow the Service Cloud, it isn’t entirely surprising to see that so many customers are focusing their use of the Community Cloud on post-sales customer service.
What I did not expect is that a larger number of Community Cloud customers are using it for partner enablement than they are for internal collaboration. Given Chatter’s roots as an internal-only communication tool, I would have expected to see more internally-focused usage of Community Cloud than what was reported. Of course, Chatter isn’t the only component of Community Cloud, but it is the oldest and most established among Salesforce customers. It will be interesting to learn more this week about why external community support is out in front of internal use of Community Cloud.

Community Cloud Business Benefits

The final area of interest here that The State of Salesforce report provides data on is business benefits associated with Community Cloud. Bluewolf compares productivity gains and cost reductions reported by two Salesforce customer segments, those who are using Community Cloud versus those who aren’t.
Community Cloud Biz Benefits
Clearly, Salesforce customers who are using Community Cloud in tandem with one or more of the company’s other offerings are realizing higher productivity and lower operating costs than customers who have not adopted Community Cloud. No surprises here. As noted above, Community Cloud is an enhancement and enabler to the other Salesforce clouds. This data is proof of that notion’s validity.

The State of Salesforce Community Cloud

Bluewolf’s The State of Salesforce report raises as many, if not more, questions than it answers about collaboration and content management among Salesforce.com’s customers. As a result, it’s hard to derive much insight from the survey data reported other than that Community Cloud is enjoying respectable adoption among Salesforce’s best customers, and they are seeing greater benefits by using it with the other Salesforce clouds, especially for external-facing use cases. While I can gather some anecdotal stories and learn more at Dreamforce this week, another survey would be needed to get the data necessary to understand how successful Salesforce’s collaboration and content management offerings have been with, and for, the rest of its customers.

Networked Business Defined

In my recently published research agenda, one of the topics that I included was networked business. I gave a brief definition of the term in that post, but realized that it would be useful to go a bit deeper. My goal is to make sure that Gigaom Research clients and followers would have a solid, base understanding of the term whenever I use it in the future.
What is ‘networked business’? Much academic work has been done to define this term in the last two decades. However, rather than forge a consensus definition from multiple opinions, I decided to build my own in 2012, based on the dictionary definition of each component of the term.
However, I quickly found that there is not a single definition for each word. Every dictionary that I referenced offered a slightly different, nuanced version of what the words ‘network’ and ‘business’ mean. In the end, I decided to work with definitions from Snappy Words, a free online visual dictionary. Snappy Words consistently offers thoughtful, refreshing definitions that go beyond the ordinary ones proposed in more traditional dictionaries. Here are the Snappy Words definitions I chose to start from:
network (n.) – an interconnected system of things or people
business (n.) – a commercial or industrial enterprise and the people who constitute it
The Snappy Words meaning of ‘business’ cited above is a good example of how their definitions are different. Most traditional dictionaries do not reference ‘people’ in their definitions of ‘business’. The central point of the Social Business movement was (and is) that people matter quite a lot in business. As is often the case, Snappy Words has done the best job of incorporating recent thought into their definitions.
Back to the task of creating a definition for ‘networked business’ from those for ‘network’ and ‘business’. Combining the two definitions was not as straight-forward as you might think. The complicating factor is which part of speech should emphasized, the adverb (‘networked’) or the object (‘business’). If the object is highlighted, the resulting definition of ‘networked business’ best applies to a single organization. If the adverb (or state) is deemed most important, then the definition most accurately describes an ecosystem
So we really need two definitions for ‘networked business’. Here are the ones that I have proposed:
networked business (n.) –  a company whose value-producing assets are connected to each other and with those of other organizations
networked business (a.) –  a state in which an interconnected system of organizations and their value-producing assets are working toward one or more common objectives
The first definition is about an individual business and the connected state it is in, internally and externally. A networked business views its organizational units as both independent silos and connected network nodes. It treats its people like individuals and co-dependent employees. The networked business sees itself as a separate entity, as well as a partner with other organizations.
The second definition speaks to the larger concept of networked business. It describes the collaborative ecosystem in which individual networked businesses work together to create and capture value. It is a philosophical objective and, if successfully achieved, an operational reality of how business is done in the early 21st century.
These definitions have held up well in the three years since they were written and first published. That said, any definition should be subject to change, as the thing that it is attempting to define morphs over time.
What do you think about these two definitions of ‘networked business’? What do you specifically like or dislike? Are there things that would you add? Please leave comments and suggestions below. Thanks!