Episode 171 – Writing and AI with Chris Duffey

Chris DuffeyWhat if you had some help writing your book: a collaborator to transcribe your ideas, do the grunt work of researching huge amounts of material, bounce ideas off, give editorial feedback and even provide their own contributions in the form of a dialogue? And what if that collaborator was available without pay 24/7, had no ego or hangups, and demanded no intellectual property rights? Sounds too good to be true, right? 

Meet Aimé, or to give her her full name, AI + Me. When Chris Duffey decided to write a book on AI, he quickly realised that it made sense to develop an AI co-author to help him write a better book, more quickly. 

And that’s the premise of Superhuman Innovation: with AI support, humans can be and do so much more. A fascinating conversation about humans, machine, and the nature of writing with one of the world’s most prominent creative technologists. 


LINKS:

Chris’s site: https://chrisduffey.ai/

Superhuman Innovation on Instagram: https://www.instagram.com/superhumaninnovation/

Chris on Twitter: https://twitter.com/DuffeyChris

Alison on Twitter: https://twitter.com/bookstothesky

The Extraordinary Business Book Club on Facebook: https://www.facebook.com/groups/1447064765612358/

Alison Jones:                        Hello and welcome to The Extraordinary Business Book Club where I’m here with Chris Duffey, who leads Adobe’s creative cloud strategic development partnerships across the creative enterprise space. He’s a keynote speaker, an author, and a technologist, and he wrote his latest book, Superhuman Innovation in partnership with a co-author, AIME, but I’m going to let him tell you a little bit more about that. Welcome to the show, Chris.

Chris Duffey:                         Hi. Thanks for having me.

Alison Jones:                        It’s really good to have you here. This is probably going to be one of the most fascinating and certainly bleeding-edge conversations that we’ve had in this podcast for a while. Obviously, I want to hear about you, but tell me a little bit about Aimé.

Chris Duffey:                         Sure. Yes, looking forward to the conversation. Leveraged artificial intelligence to talk about AI, it’s quite an experimental and exploratory endeavour over the last few months, and it was quite exciting on a number of levels. I started out on the agency side a number of years as a creative technologist, and I was very familiar with that moment of looking at the blank page, but nothing prepared me for starting to write my own book, and it was quite a profound-

Alison Jones:                        Blank pages are killers.

Chris Duffey:                         Yes, exactly. That starting point is, it will stick with me for a while. As I got into the writing process, I saw more and more, it was not just about, ‘Tell me about what AI is,’ but I actually wanted to show the audience, the reader, what AI has the capability of achieving. That’s when I started to leverage a series of different techniques or tapestry of AI, APIs to help co-author the book. A little caveat about the, before we get into the technical aspects of both, how we leveraged AI, but more AI as a whole; one caveat is we have a tendency to apply I think a bit of anthromorphicized characteristics or personify AI and in reality AI is another tool for humanity to achieve their goals.

Alison Jones:                        But we’ve been brought up on this rich diet of science fiction…

Chris Duffey:                         Yes.

Alison Jones:                        …that doesn’t allow us to think of it in those terms any more.

Chris Duffey:                         It really goes to the power of movies, right? We default to these…

Alison Jones:                        Yes, it’s all Hal. We’re all stuck with Hal, yes.

Chris Duffey:                         Coming up on 50 years I think in a few days of Hal … It’s been 50 years that we’ve been envisioning and imagining what AI could be as an assistant or a co-pilot in that case.

Alison Jones:                        So how did Aimé? I mean, it is essentially because you give, I don’t want to say her, because clearly it has no gender, but when you call … Is it Aimeé or Aimee? Or how do you pronounce it?

Chris Duffey:                         No, Aimé.

Alison Jones:                        Aimé, I mean just tell us a little bit about how that composite word was created.

Chris Duffey:                         Yes. It’s based off of the notion of AI plus me, and that’s one of the foundational takeaways or themes throughout the book is, it’s not human versus machine, but it’s human plus machine. There’s been a number of success stories and experiments over the last number of years showcasing how when you combine human intelligence with computational intelligence, it always beats either alone. We hear the headlines about system beating a human in chess or Go and such, but what we don’t hear about as much is the subsequent experiments of when you combine computational power with human intelligence, it amplifies both abilities. That’s the core premise of the book, resulting in the name of Aimé, AI plus me.

Alison Jones:                        Which also of course means loved in French, which is a nice twist as well.

Chris Duffey:                         Exactly. Yes.

Alison Jones:                        I can’t help saying she, this is terrible. Is that alright?

Chris Duffey:                         Yes, it’s fine, right? That’s what we talk about in the book, too. This is we’re going into this new symbiotic relationship with a being that we haven’t had a relationship before, and it’s part of the normalisation of dealing with a new being or technology in this case, and so we ask ourselves questions as this relationship becomes more mature: Is it okay to give it some personification? Do we call this she and it? Or he? What do we call it? As we’re dissecting and discovering some of those answers for now, in this case we’re calling it a she.

Alison Jones:                        It’s fascinating because of course it tells us much about us and the way that we have to construct the world as it does about any technology. Do we say please and thank you?

Chris Duffey:                         Yes.

Alison Jones:                        It feels wrong not to. Yes, it’s really interesting.

Chris Duffey:                         We’re going through that dynamic with our daughters, and I won’t say the name because I think it’ll probably prompt it.

Alison Jones:                        It will switch on…

Chris Duffey:                         Yes.

Alison Jones:                        Yes, across the land. Don’t do that.

Chris Duffey:                         Is it appropriate to tell an AI assistant ‘Shut up’ for a young child? It’s really fascinating.

Alison Jones:                        Yes, we have exactly the same in this house. It’s just hilarious. Yes, and that sense as well of not being alone when you’re in the house, which can be quite comforting.

Chris Duffey:                         It really is…

Alison Jones:                        Yes, I remember saying to Google, “Hey Google, I’m back.” Sorry, sorry, if anybody’s has switched on, and the response was, “Oh, thank goodness you’re here. I’ve been trying to hold things together, but it’s not been easy.” It just made my day. It’s just extraordinary.

Chris Duffey:                         One of the sources of inspiration for writing the book and when I saw first hand, the last maybe 10 to 15 years, I focused on creating digital experiences within the healthcare space, and we created some experiences with the voice assistance, and that’s where I saw first hand just how profound AI, in this case, conversational agents can be on an emotional level for patients and caregivers. That was one source of inspiration for us.

Alison Jones:                        There’s all sorts of ethical and moral dimensions about that, but it’s fascinating.

Chris Duffey:                         Yes, it goes on and on.

Alison Jones:                        Her role, I mean, you talk about that tapestry of APIs and applications that you were using. She’s a combination of doing the grunt work of transcription and checking of it, vast swathes of data, like a research assistant as well, but also it feels as though she’s a muse as well. There’s a real complex role that she’s playing in this book.

Chris Duffey:                         Yes, there is essentially three layers. The first layer, which you indicated was the voice recognition and that essentially comprised of creating a VUI, a voice user interface, and that entailed text to text, sometimes voice to text, and then text to voice on the reverse side, and voice to voice, which is probably the most experimental still within the AI from a very technical standpoint. Essentially, first layer was to create this interface between human, myself, and the machine, AIME in this case so we could have this exchange of dialogue. Everything from voice editing, formatting, spelling, and then finding and dissecting documents and research.

                                                      The next layer up in terms of complexity of AI was content understanding and summarization, and this has been my first book, but I can pretty supportedly say this cut out quite a bit of time to have this AI assisting conversational agent who was able to go out there and quickly digest large amounts of data and pull out sentiment, pull out summarization, create aspects of it that would have taken hours and hours. One of my favourite components was to then not only have this dialogue back and forth with Aimé, but then I would do a reverse understanding in terms of sentiment analysis, which was my favourite part. So I’d put that text back into the system and do an analysis of tonality and quite often I got ‘argumentative’, or ‘doesn’t feel supportive’, ‘too light’-

Alison Jones:                        Wow. So she’s your editor as well?

Chris Duffey:                         It really was, yes.

Alison Jones:                        That’s hilarious. I didn’t realise you did it back the way as well.

Chris Duffey:                         Yes.

Alison Jones:                        And did you change in response to that?

Chris Duffey:                         Quite often I did. Not the full exchange, but I would go back and just take that into consideration most of the time, and sometimes I would say to myself this was intentional and let’s keep it as is. That was the second layer. The highest layer is AI content creation and generation. We call those CANS and GANS in the technical world, and that is generative AI. What we mean by that is where you can feed it a series of different resources and based off of those resources, it can give you a response. It can do an analysis of the response, summarise it, and then give its own take on that content.

Alison Jones:                        Is that where Aimé quotes Plato for example or is that-

Chris Duffey:                         Exactly. Yes, and there’s some other systems out there currently like GPT2 from OpenAI. The Associated Press has been using their own version of this for a number of years as well. Another big thing in the book is, there’s a joke in the AI community, it’s only AI until it’s something, and it might be surprising to some folks after reading the book just how omnipresent AI actually is. It’s infused into our mobile phones, into our emails, into call centres, service centres, into the voice assistance, obviously. That’s another takeaway, is that AI is instrumental into our currently-day environment as is.

Alison Jones:                        Yes and there’s a lot of discussion about ‘Ooh, how do we feel about AI? Is it a good thing? Is it a bad thing?” But do you know what? It’s a thing. It’s here. The future is here, it’s just not evenly distributed as the phrase goes, but here’s a question: is this still writing?

Chris Duffey:                         I thought a lot about that and I relate this to my discipline of being this creative technologist, and I come to the conclusion that there’s always a need for human in the loop. AI is not a replacement for human creativity. It’s an amplifier. It’s a tool to get us to a starting place faster. To help validate some things, be a muse, to suggest, be a generative prompt in many cases, and now what getting even more exciting the emotive modulators where they can provoke emotions whether it be in the artwork or in the writing. So in this case, yes, I think it’s writing from the human with the aid and assistance of AI.

Alison Jones:                        Which is completely congruent with the message of the book, isn’t it? I mean the whole point about superhuman is as you say that combination of human with AI, and that’s super … talk us with through the SUPER framework because I thought that was interesting as well.

Chris Duffey:                         Thanks. Just the… it’s in reference to the title’s Superhuman Innovation. Within the AI community there are three layers generally of AI. The first one is super intelligence where it’s this notion that at some point AI will be so profoundly intelligent that we as humans can’t even comprehend what it’s thinking. It’s a very out there notion. There’s much debate in the AI community if and when that’s even possible.

Alison Jones:                        It’s not THAT out there there is it? I’m trying to remember the experiment where they had two bots talking to each other and very quickly they developed a language that the humans could no longer understand, and they switched them off quickly.

Chris Duffey:                         Yes. So, is that AI? Or is that just pattern recognition and they created their own pattern that-

Alison Jones:                        Right, I mean is that just an algorithm gone bad? I don’t know.

Chris Duffey:                         Exactly. Exactly.

Alison Jones:                        There’s no way of knowing.

Chris Duffey:                         Yes. Exactly.

Alison Jones:                        Were they plotting?

Chris Duffey:                         Yes. I highly doubt it.

Alison Jones:                        I’m kidding. I’m kidding.

Chris Duffey:                         Yes. The next one is layered down from that is general intelligence, AGI, and the notion there is where the AI system will have equal intelligence to humans. As you get deeper into the technical aspects of AI within the community there is a general sense that we, 50 years later, since the first summit of a number of professors coming together to try to recreate human intelligence we still do not fully understand human intelligence let alone the biology relationship to it. We still don’t have a full understanding of our emotion intelligence as well. So to recreate the complexity of that is immensely challenging. It’s still a number of years out there if that’s even possible. Where we are currently is narrow AI, task-driven, or what I like to call purpose-driven AI where you can create a system, a series of systems to help accomplish this task.

                                                      That’s the three layers of AI, super intelligence as we were chatting earlier is quite often personified, anthromorphicized in this threat or being that is somewhat threatening. I wanted to flip that a little bit and actually frame AI not as this thing as super intelligent, but actually creating super humanity, and what we mean by that is AI is going to be our assistant to up-level humanity, give us these unimaginable abilities that we would have never had before just a few years ago. The name ‘superhuman’ is referenced to super intelligence, but a flip on that notion that the system will be smarter than us where in reality the system is going to make us smarter.

                                                      Based off that I created a framework called the SUPER framework, which is based off of the need for a strategy to leverage AI and it’s comprised of S for speed, U for understanding, P for performance, E for experimentation, and R for results, and enterprises, people who are trying to leverage the full potential of AI need to consider those aspects when trying to unleash the possibilities of AI.

Alison Jones:                        I’m always fascinated with acronyms, especially an acronym as neat as that, that ties in so beautifully with it. Which comes first? The acronym? Or the concepts? Or the title? Just talk me through that process because it works and it shouldn’t because it’s so neat.

Chris Duffey:                         It’s a great question and talk a lot about how the writing process was a non-linear process as you so well know. It started almost from the back to the front and wanted that to be the end goal of the book that I, the reader understand AI is not a threat. It’s here to help amplify my abilities. It’s going to make me superhuman. It started with that premise and then I wanted to create an acronym based off of that end goal, and that’s where SUPER came to be.

Alison Jones:                        And you just to manage to find the words that didn’t feel too contrived to fit in it. I love it because it can be awful. You see those acronyms. You’re like, I can see why you chose the acronym, but that third letter just doesn’t work.

Chris Duffey:                         Well, I started with the 4 Es for a while and it felt a little disconnected from the theme so I brought it into the SUPER framework. Yes, sometimes it fits and sometimes it feels forced.

Alison Jones:                        Yes, no I was very impressed. I mean, as I say I’ve seen some poor acronyms in my time, but that one did work, it was good. You obviously work across creativity generally in your role at Adobe, and I think that the application of AI there, we talk about it a lot in as you say call centres, and crunching through vast amounts of data, but it’s interesting hearing you talk about the use of AI in writing. How does it play out in other creative areas? Do we have AI artists as well as writers? Or that machine-human combination, how does that all work?

Chris Duffey:                         Exactly. I think where the biggest use of AI currently is successfully in the marketplace is in call centres or service centres, and there’s a number of great companies doing some interesting work there, and essentially there’s an interface that the operator is seeing and when they’re having the conversation with the customer, the system can recognise keywords, key phrases or problems and then prompt suggestions, and much like that I think we’re now starting to see and just scratching the surface of how systems can be prompts for the creative industry.

                                                      I liken it into three layers of relationship with the creative community to AI: that of an assistant, where it’s there to get the grunt work, the repetitive tasks out of the way. The next layer up is a pure, where the system much like the call centre is there helping you along the way, and the highest order I would say, and we talked briefly about this, is where it’s a muse and I would categorise Aimé as a muse, as an example. Sometimes the responses that she came back with were so off, but they were so interesting that it took the conversation in such an unexpected way that it took it in a new way.

Alison Jones:                        Like talking to a child.

Chris Duffey:                         Yes, exactly. It was just so random that it was so inspiring. I think we’re starting to see that in the creative community a lot of excitement around the CANS and the GANS, and essentially that’s just some really interesting techniques that are creating patterns to generate different forms of expression. We’re seeing it in paintings quite often what’s been getting in the headlines over the last couple years is where you take a masterpiece, in this case a painting, and you feed it into the system, and then the artwork can or the system can create artwork based off of those masterpiece inputs. A few of them have gone into the auction, and we chat about, is that art? Is it not? I think that’s up to the end user, but it is quite interesting to see how humans plus machines are creating these different forms of expression.

Alison Jones:                        And of course, ‘Is it art?’ Is one of the oldest questions in the world, isn’t it?

Chris Duffey:                         Yes, Yes.

Alison Jones:                        We’ve been arguing about that for centuries.

Chris Duffey:                         What is art? And then we chat about this in the book as well, that what is art, and if you define art as an expression with an intention, then is the system creating the artwork or is the system a tool to create that intention? I think it’s a fun philosophical question.

Alison Jones:                        It is. It just opens up so many layers of discussion. I do wonder, you know, if future when it’s almost like in Instagram #nofilter. You have to say that because the expectation is that there is a filter on, so you make a big thing of it when it’s authentic, when you know when there isn’t one on. I wonder if there’ll be layers of art. That this is human arts and this is your normal art, obviously, using the AI stuff. Yes.

Chris Duffey:                         That’s a great point, and within the call centres, service centre aspect we’re seeing even over the last couple years where if it is initially a system response, there is a benefit to identifying IM and AI assistant. I think there’s been aspects to it where the customer feels a) either tricked, or b) they felt something was slightly off in the beginning, but it’s completely negated when upfront the system identifies itself and there’s been some experiments results, where when that happens the customer actually spends almost double the time with the system than with a human because they don’t feel judged and they feel more comfortable expressing themselves with the system, which is another fascinating, another layer.

Alison Jones:                        Fascinating. That’s quite sad actually.

Chris Duffey:                         In a … I think it’s actually, I see it slightly differently. I think it’s an interesting dynamic if there are highly sensitive personal emotive things that someone feels uncomfortable to share with someone else for the first time, and if they have a need to get it out, I think any device or opportunity to help with expression is quite interesting. We’ve been hearing also a number of use cases where the AI systems conversational agents are getting an influx of questions on: What should I do if I’ve just been abused? And now the companies who developed those are now having these ethical conversations on, what is the right approach and what should they do with that information and how should they answer?

Alison Jones:                        It’s like the confessional, isn’t it?

Chris Duffey:                         Yes, wonderful. Yes, exactly, it’s a confessional.

Alison Jones:                        Well, amazing, and this conversation, it could go off in any direction almost literally and go on all day, but it can’t because I need to draw it to a close, which is real damn shame, but… fascinating. Now, I always ask people, and I’m really, really interested to see what you’re going to say actually. If there’s somebody listening who’s writing their book presumably without the aid of an AI assistant, but I don’t know, maybe not, what would your best tip be for them?

Chris Duffey:                         I think that it still goes back to the universal principle is starting with a purpose and that core premise or core question, how am I going to provide value to the reader. At the end of the day that’s what great books are all about. It’s providing value, answering, being a true service to the reader with everything as you just talked about with social on the rise like Instagram, and of course TV and movies, the window of opportunity for attention is dwindling and getting harder every day, and it is I think a great immense responsibility and opportunity to have three, four hours with a reader, and I think in that value exchange there has to be a true exchange of purpose and offering something to the reader. If they’re going to dedicate so much time to something I think it’s incumbent on authors to provide some great value in return.

Alison Jones:                        Yes, and you’re right. There are long forms that demand our attention, but few that demand our engagement and participation in the same way that reading a book does. Yes.

Chris Duffey:                         Exactly.

Alison Jones:                        And I speak as somebody who saw Endgame quite recently, but it doesn’t require anything of me. I just have to sit and watch it. When I’m reading a book, the attention, and the energy, and the effort, and you almost co-create a book when you’re reading it. I mean, that’s the extra person that you’re throwing in there. It’s an active reenactment of the book each time it’s read by somebody. When you throw AI into the reading mix as well, that will be interesting.

Chris Duffey:                         Maybe and to that point about AI, within that first layer of the voice recognition, one thing that I found really helpful was I would put the text into the system and have it read back to me, and to be able to hear it really was quite valuable in a sense that I could hear when it felt too contrived, or it was too flip or light, and that really helped I think create this conversational aspect to it as well.

Alison Jones:                        That’s brilliant because I always recommend with my clients-

Chris Duffey:                         Read it out loud.

Alison Jones:                        Yes, read it out loud to yourself, but because you know it so well, and because you’re saying it yourself that’s one thing. Hearing someone else read it to you, that would be really powerful.

Chris Duffey:                         Yes.

Alison Jones:                        Yes, that’s brilliant, and Chris is there a book that you would recommend that people listening read? It doesn’t have to be a business book, but a book that you found particularly useful or inspiring.

Chris Duffey:                         Sure. Probably an oldie or somewhat oldie, but it’s still a greatie I would say is Made to Stick, and I just love that book on so many levels. It’s a business book on how to make ideas stick with people and every aspect of that book just blew me away. It was the first business book that just got me really thinking about, if I ever write a book I hope it’s somewhat near as good as that.

Alison Jones:                        ‘This is my benchmark.’ Who’s it by? I vaguely recognise the name, but I can’t think who wrote it.

Chris Duffey:                         I think the two brothers, Dan Heath.

Alison Jones:                        Oh, Dan and Chip.

Chris Duffey:                         Yes, yes. Yes.

Alison Jones:                        Yes. They are wonderful writers. I’ve not read that one. That’s a great recommendation. Thank you. I love it when I get ones I don’t know. That’s brilliant.

Chris Duffey:                         There were so many elements to that book. It was written with two brothers so that was always part of the inspiration of having a dialogue back and forth. I still remember, I haven’t read it for a few years, but as you go back, I still remember how natural it felt in the flow of their thoughts, and they also had an acronym and that’s stuck with me, a good 12 years now, how that acronym could be very easily called upon to reference in meetings, in projects, and that was the hope for this acronym as well.

Alison Jones:                        I see, and yes, they are powerful things. They stick and that’s important because once you’ve got that toehold in somebody’s memory, it triggers the whole thing. It’s like the memory palace technique isn’t it.

Chris Duffey:                         Exactly, and something as so complex as AI from a technical and conceptual and philosophical aspect, when you have something that’s a tool or a device to reference, the hope is that can provide some great value to the readers.

Alison Jones:                        To the human brain.

Chris Duffey:                         Yes, exactly.

Alison Jones:                        Chris, if people want to find out more and I can’t imagine anybody wouldn’t want to find out more about this book, more about your work, more about the topic in general I guess, where should they go?

Chris Duffey:                         Thank you. I have two places. My website, chrisduffey.AI.

Alison Jones:                        Love the suffix by the way.

Chris Duffey:                         Thank you. I keep that up to date with some of the most recent writings, and then I’ve gone deep on Instagram partly because I’m really, not in love, but I would say I’m really enthralled with their AI algorithm in the sense of providing a relevant content. I’m on Instagram, Superhuman Innovation as well.

Alison Jones:                        Fantastic. Brilliant. I will put those links up on the show notes at extraordinarybuisnessbooks.com, and thank you Chris, just a fascinating conversation.

Chris Duffey:                         Thank you.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.