Episode 92: A Conversation with Chris Duffey

In this episode, Byron speaks with guest Chris Duffey about the nature of creativity and how machine intelligence can interact with creative concepts.

:: ::


Chris Duffey is author of Superhuman Innovation: Transforming Businesses with Artificial Intelligence, and the Head of Artificial Intelligence Innovation and Strategy at Adobe. Duffey spearheads Adobe’s Creative Cloud strategic development innovation partnerships across the creative enterprise space.


Byron Reese: This is Voices in AI brought to you by GigaOm and I'm Byron Reese. Today my guest is Chris Duffey. He spearheads Adobe's Creative Cloud strategic development partnerships across the creative enterprise space. His keynotes have received over 50 million impressions. His work has been featured in a hundred or more global media outlets including The Wall Street Journal, Inc, Adweek and on and on and on. He's also the author of Superhuman Innovation: Transforming Businesses with Artificial Intelligence. Welcome to the show, Chris.

Chris Duffey: Thanks for having me. Great to be here.

I'd like to start off with: what is intelligence? How would you answer that and why is artificial intelligence 'artificial'?

Great question. Love that question. You know specific to artificial intelligence, I always default to the definition that AI is the study and design of intelligent agents that are able to not only perceive their environment but also act on it. Maybe a higher level definition of just human intelligence.

And it's interesting that there's some theories that it can be defined in nine categories or nine different aspects of intelligence: that of intrapersonal, that of spatial, that of naturalist, musical, logical or mathematical, existential, interpersonal, bodily and then linguistic. So those are kind of the two definitions that I refer to for both AI, as well as human intelligence. Even maybe a third aspect of that is—like the old quote "the true sign of intelligence is imagination" and that's where I have a personal fascination of this intersection of artificial intelligence to help amplify human intelligence.

So you said “perceive their environment and act on it.” Would that go all the way down to say, a cat food dish that can tell when it's empty and then refills itself?

I think that would be a very mechanical example of a tool that is being used to enhance...

Is ‘perceive’ really the right word? Does a computer perceive anything or does it just measure things?

Yeah, I think that's an accurate word exchange.

Because it seems like we use all these words like the computer ‘sees’ this, the computer ‘knows’ this, the computer ‘figures out’ that. I think we use those words because we don't have equivalent mechanistic... we're not used to machines doing what computers do, so we have to use these human words.

But do you think one runs the risk that somehow gives people a heightened sense that the computer really is a thing that has a ‘self’ and all the rest?

Yeah, I think we do run a tendency to almost anthropomorphize artificial intelligence, whereas you know at the end of the day it's a machine. We are biological humans. So you know maybe there is some ‘self projection’ for lack of better words, or better categories to apply to them. But I agree they are machines, mechanical entities that sometimes we apply human qualities to because of a lack of other reference points.

Well if we're gonna use the term ‘machine’ to mean like a device that is completely governed by the laws of physics(because you know it's deterministic, you crank it up and it runs), do you believe that people are machines? Are our brains just machines?

Without a doubt we are biological creatures. So by no means are we machines. I think we're completely two different entities. We create the machines and we are biological, so I think there is a very distinct difference in my mind.

But you know the old argument that if you figure out what a neuron does and then you reproduce that and then we figure out what every one of your neurons is and we build that, I mean at what point could we not build something that when you turned it on, is you, or identical to you?

You know I think with the course and history of artificial intelligence, there's been this tendency to recreate the human mind. Whereas I'm of the mind where AI should be purpose driven and task driven, so more of a narrow AI viewpoint on things rather than a general AI aspect to things.

But back to your reference [to] the neurons. I think that there again we're talking about biology, and biology can be somewhat cloned or duplicated in a biological environment, but we're almost playing with different ingredients. If someone is trying to create an AGI using computational sources, again I think they are different distinct entities.

So when you made that comment a minute ago about narrow AI vs. general AI, were you saying you don't think we can build a general AI? Or do you think the most useful AI is a narrow AI?

For modern day businesses where my focus is, narrow AI, without a doubt, is I think the game changer of our generation and generations to come. So my focus is on how can we implement and create greater absorption of narrow AI. I'd like to refer to it as purpose driven AI or human centric AI.

So you talk about AI and creativity and innovation. Can artificial intelligence be creative, and what is creativity, for that matter?

Yeah. And this is where I really love to explore, and it's much of a similar conversation in that we have of: ‘Can AI replicate the brain, the mind, consciousness, human spirit?’ and you know another subset of that is creativity. It's fascinating in the sense, when you talk to creatives (and I came up through the ranks as a creative professional), unlike I think in the scientific realm, where there is a lot of reflection on what is intelligence, artificial intelligence. There's a slight nuance in the creative community where, although we do take time to take a step back and do self analysis of what creativity is, I think though the nuance there is we spend a greater amount of time just going out and being creative, designing things, creating things, being inventive. So that's where those two aspects kind of fork off in the realm of creativity.

But to define creativity, again is an elusive almost ‘suitcase’ word similar to artificial intelligence. Very simply put, creativity is about connecting things that are seemingly unconnected from a first pass. So back to your question: ’Can AI in this case, be creative?’ I think at this point, AI is very good at automating certain aspects of creativity—that repetitive mundane repeat task. And I think within the creative AI community, there is a spectrum of feelings of how high AI can get in terms of augmenting and generating as well.

So if creativity is about connecting things that weren't obviously connected, help me understand how J K Rowling's Harry Potter series or Lin Miranda's Hamilton or any of those, or Banksy's graffiti... what unrelated things are they connecting to make those works do you think?

Yeah. Imagination is essentially taking creativity and connecting these seemingly unconnected things together, and you know there's a number of different levers that can be pulled in terms of visual aesthetics, emotions, even music for instance. So AI can in this instance, up level humanity to be more creative and I think right now, we're at that level of helping creatives on a professional front to get [those] repeat mundane tasks out of the way so they can do these higher level creative aspects of creative generation.

There's one notion that creativity can be broken out into three aspects: that of bending, breaking and blending. And I think this gets to your question [about] what aspects of some of those artists, what they were using, and you know I think artists kind of weave in and out of those three elements on a very fluid level. From a bending aspect, they take something that kind of exists and slightly bend it to create something new or unexpected. From a breaking standpoint, you know it's taking something that hadn't existed before and just kind of broke it and re compiled some parts from different things. And then the blending, which I think there is a big natural focus is to blend these different worlds into something new.

You know it's really funny, did you used to watch Seinfeld back in the day?

Of course, of course.

And you know the cliche about it, the joke supposedly is: it's a show about what?

About nothing.

That's what they say. But what's really interesting is it was pitched as a show about where comedians get their ideas, and that's why you always have Jerry doing a standup routine about whatever the show is about. And so I guess that question of: where do ideas come from? Will they come from computers do you think?

You know there's maybe a couple aspects to that question. Right now I think the creative output (and I make that distinction ‘output’ rather than creative systems), is being generated greatly. A lot of recent excitement around the CANs and the GANs, those creative outputs are coming from derivative... there's a derivative aspect to it. So in a lot of experiments are examples of sourcing or referencing masterpieces. I think that's greatly due to... those are visual mnemonics to signify very quickly like: ‘this is a symbol of art.’ So they're referencing you know symbolic pieces of art and then they're having almost this generative aspect or interpretation out of those sources. So there's a derivative aspect to current day systems.

There's much debate [about whether] AI will truly be able to be creative on its own. And depending on the day I go back and forth. As of now I firmly believe creativity is fundamentally a human characteristic. Yes, AI systems can help generate, can help be almost a muse for creators, but at this point and for the foreseeable future, systems can't have intent to be creative. That's where I kind of define or draw a distinction between creative output and creative systems.

It's interesting because you remember when Lee Sedol was playing AlphaGo and AlphaGo beat him 4-1. There's a point at which there was just one move that’s called historically Move (whatever) 37. I don't actually remember the number where the AlphaGo team said they were like ‘whoa where did that come from?’ And they looked into a human player according to their database, there's only a one in thousand chance a human player would have ever made that move, and that was kind of the first moment that people started talking about AlphaGo's creativity... that it played the game creatively.

And you referred to creativity as a ‘suitcase word’ and maybe it's not even a useful concept. I have this sneaking suspicion that these words that are so hard to pin down, maybe that's because they don't really exist. Maybe there really isn't something called intelligence that’s useful, if we can't even say what it is. Maybe there isn't really something called creativity. And we maybe need a different word for it. So would you even, would you go that far? Or could a super inspired move in a game of chess or AlphaGo or something be creative?

You know, back to your earlier example about perception. What do machines really perceive? I would almost apply a similar lens to that creative move. I would almost gravitate towards ‘unexpected’ or... yes, unexpected. I would gravitate towards [that] rather than creative.

Fair enough. I guess it's the distinction of: are they creative or are they mimicking creativity or is there a difference between those two things? So your new book is called Superhuman Innovations: Transforming Businesses with AI, What caused you to write it? And how do you transform? What's the general thesis?

Sure yeah. Thanks for bringing that up. You know I came up through the ranks as a creative in the ‘net, the last number of years. I was on the digital health side helping pharmaceutical companies and biotech companies not only promote their products, but also I've helped create some programs for patient centricity and patients. So over the last number of years, worked on maybe three or four initiatives that really opened my eyes to the power of artificial intelligence in terms of human health. We worked on some where we were using AI as a voice assistant for patient care. What we saw there was patients were having almost this unguarded ability to talk with virtual assistants because they felt they weren't being judged.

Another example was we are working on orphan disease states, these diseases that aren't that recognizable in the population or aren't seen that often by physicians. Then we had this insight or observation that physicians and the patients were sometimes going undiagnosed for upwards of seven to 10 years with these awful symptoms. And so we saw that this opportunity to leverage AI to help identify these disease states, and we saw some great success there as well.

And then we also did an initiative where we were measuring, through AI, the ability to measure atrial fibrillation via the camera on a smartphone; and then we overlaid AI on a second level to try to extract predictions on lifestyle, demographic or geographic aspects from a broad population standpoint. So that really got me quite excited, and the deeper I got into AI, the more I saw the need for this, what I call a ‘middle layer demystification’ of AI.

I think there's some great thought leaders’ books out there that are very philosophical, and I think they gravitate towards: what is human intelligence? What is consciousness? And I think there's a time and place for that. It's very important, but it can get very philosophical. And then there's this other layer that's very technical, and of course we need to understand the technology. But for a business application, there's very few I think resources that talk about ‘what is AI? What is it not? How can I apply it to my business?’

And once I got into the process of writing the book, I had this almost ‘aha’ moment where there is this need not only to talk about AI, but actually leverage AI in the content creation process of writing the book. So that's where, within Superhuman Innovation I leveraged an ensemble of different techniques to help co-author or help write the book. And to your point: for lack of better words or descriptions, we apply sometimes human words to AI, and I think there's kind of a conversation around ‘Can AI be a co-author?’ but for lack of a better word, it was a ‘creative assistant’ for helping write the book.

Long story short: the premise of the book is to be a rallying cry for businesses, organizations, whether big or small to see the opportunities in leveraging AI within their organizations. A couple stats that always pop to mind: by 2020 companies that are using AI will take $1.2 trillion away from companies that don't.

Where is that from?

That one is from M.I.T. School of Business. And being a creative, I think much like... some of these stats, I think we probably have to go in and really delve and dig into how they come to those.

I'm not doubting, it sounds right. I'm just curious about like the methodology. But I mean I think I interrupted you. What was your next stat?

I think the long story short with all these stats are it's a huge opportunity for businesses to start to implement AI within their organization. So within Superhuman the fundamental premise is: AI is not here to replace jobs. It's here to up level us to do a higher level of critical thinking. It's a true competitive advantage for organizations to leverage going forward.

I agree with that. I write extensively about it and so forth. Why do you think there's one camp who says “Wow these technologies increase human productivity and that's always good—more productive humans are good. If you don't think that, then somehow lowering our productivity would be good for us and that seems farcical.” And so you've got these people (I count myself among them) who say: “You know, therefore these technologies can't really cause unemployment, they just increase people's productivity. And that's where all economic growth comes from. We don't live better lives than our great grandparents because we work harder than they did, far from it, but because an hour of our time is just so much more productive.”

But then there's a whole school of thought that says ‘You know there's a whole other group of people, also in the industry, who say you know, don't kid yourself. Many of the tasks that make up many of our jobs, machines will be able to do and do better than humans. And you start with order taker in a fast food place, you replace that with a machine and you just start doing that en masse. And eventually there really is a shortage of jobs.’ So knowing both of those positions, everybody's got an opinion. But why do you think people in the industry differ so sharply on this question?

Yeah, so my focus background is creativity, content creation for innovation and ultimately business transformation. Within that industry if people agree or can agree or just understand the premise that human creativity is profoundly human, let's see how AI can help in that aspect. So what we're seeing very early on is (and we're just scratching the surface with AI in terms of this specific industry) it's helping automate, so it's getting that repetitive repeat mundane tasks out of the way.

And I can guarantee pretty solidly that creative professionals who got into the industry and studied through formal traditional training or just got into it through apprenticeships, I can guarantee they got into the creative realm because they loved higher level critical thinking. They didn't get into it to do these repetitive tasks.

So I guess the counterpoint to the school of thought that's taking the stance that AI will through automation, take away a number of jobs, I would reference to this example where it's taking away low level tasks, but it's not taking away jobs and inversely, I think it's creating jobs. It's creating up level learning in this instance of creativity and innovation.

I think of it in a similar way. I mean to me, if you had gone back in time 25 years and told people about the Internet and how many people would use it, and said “What's that going to do to jobs?” It's easy to see what it's going to destroy. Yellow Pages and print newspapers and mail catalogs and stockbrokers and travel agents. But you never could see what it would create. I mean we just can't [have imagined]... you know Uber, Airbnb, Google, Amazon, Twitter, Facebook and all the rest. So it's so easy to see one side of the equation, but we we lack imagination to see the other side.

I completely agree. And I think that's where we're seeing, over the last 15-20 years, we've been talking a lot about ‘the digital transformation,’ but now we're seeing true business transformations. You know every 10 to 15 years we get this new technology that fundamentally changes things. Maybe 30 years ago we had the desktop revolution and that really helped democratize content creation. Fast forward maybe 10 to 15 years, we had the mobile revolution, where now we have the ability to access all of this great content.

And over the recent years because of all of that digital exhaust, we saw the need and the rise of cloud computing to house all of this data. And so now we're at this almost profoundly ripe position where we have this surplus of data, referencing back to those examples within the health world, but it goes across all industries.

We have this surplus of data and now the true business on lock or business transformation opportunity is to extract insights for greater experiences, and talk a lot about the need to move into... we're really living in this era of the ‘experience economy’ to create these ‘one to one’ very personalized experiences with businesses and consumers or even humans for a better word. So that is the true opportunity for businesses, societies and culture: to leverage AI to create these better, more informed, intelligent experiences.

So final question, I assume just from chatting with you, you're optimistic about the future of our planet because of this technology. Or am I putting words in your mouth?

With those words, please do. Yeah, without a doubt... quite excited about the immense opportunities with AI across of course, all industry sectors. Specifically to your earlier point, the aspect that AI is going to ‘up level’ humanity in the sense that it's going to allow us to do this higher level thinking that's going to usher in this age of innovation, which I think is quite exciting.

So Chris Duffey, how do people follow you? How do they keep up with all the stuff you're doing, buy your book, Superhuman Innovation? But what else can they do?

Yeah. Recently I've been getting pretty deep into Instagram, so that would be the preferred channel at the moment. ‘Superhuman Innovation’ on Instagram.

All right. Well I want to thank you for a fun wide ranging chat for half an hour, and I hope you'll come back some time.

Thanks for having me.