Transcript: How Publishing Can “Shimmr” With AI

Interview with Nadim Sadek, author, Shimmer, don’t Shake

For podcast release Monday, October 2, 2023

KENNEALLY: “Tracing the histories of publishing and AI, Shimmer, don’t Shake argues that despite risks of negative disruption, responsible implementation of AI aligned with publishing’s core creative mission offers valuable opportunities, and the industry should embrace AI’s next wave through evolution, not resistance, to allow human and artificial intelligence to complement each other.”

At least that’s what Claude says – Claude, the AI chatbot that can digest a report or a book, then summarize it and answer questions. Shimmer, don’t Shake human author Nadim Sadek lets Claude the machine write the closing chapter of his new book from Mensch Publishing, distilling the work if you don’t want to read it all.

Welcome to CCC’s podcast series. I’m Christopher Kenneally for Velocity of Content.

In Shimmer, don’t Shake, Nadim Sadek entwines the histories of AI and book publishing to make his case that AI technologies can complement, and not necessarily replace, human authors, editors, and publishers.

Sadek, a serial entrepreneur who has worked in fields as disparate as market research and whiskey distilling, has also just launched his latest company, Shimmr.AI, a service that produces a unique DNA print for any book. Shimmr’s book DNA becomes the catalyst for online advertising aimed at expanding the book’s reach and resonance. Nadim Sadek joins me now from London. Welcome to Velocity of Content.

SADEK: Thank you very much. Goodness, it feels like you’ve been inside my head and understand everything I have to say. So it’s a pleasure to be here, and I hope you can add something.

KENNEALLY: Well, we will have to test that. At least Claude understands you. We know that much. Shimmr.AI, your new company, has what you call a noble mission – to help more human genius to shimmer in the public’s eye. It sounds like a kinder, gentler digital disruption than the upheaval that many in publishing say AI poses to their business and to literature.

SADEK: You know, when I hear that I’ve used the word noble, I sort of shudder and shake at my own shimmering. It’s a commercial business. Shimmr is set up to make money. So let me not disguise that or conceal it in any way.

However, I love that what we’ve alighted upon as a business is the surfacing, the celebration, the kind of joyous manifestation of human genius, which I think too often lies shadowed in the recesses of backlists of publishers. People have been commissioned to write books. They’ve authored them. Publishers have given their editorial teams time to toil over them. They’ve eventually been printed, bound, packaged, distributed, retailed – unfortunately, sometimes returned, sometimes pulped. There’s a huge investment in the exhibition of human creativity that just languishes in the dark too much, and I think it’s just because it’s been very hard to make it obvious to people that this stuff is available and that you can access it and enjoy it. So if there is a nobility to our purpose, it is the assistance of that creativity being available, coming to people’s notice, and actually ultimately matching it with their psychological profiles and preferences.

KENNEALLY: In your new book, Nadim Sadek, you imagine AI and publishing interacting courteously like diplomats. Your own DNA has a diplomatic strain. You identify as half-Egyptian, half-Irish, and you were raised around the world in what you call a United Nations family. How does that life experience offer a path forward for publishing as it enters the age of machines?

SADEK: I think if there has been a beneficial effect of growing up around the world in many cultures – and I feel I’m very privileged to have been exposed to so many different influences in life – I think it is that it makes one more tolerant. I think it is that one hears with more beneficence people’s different points of views, even if they’re in great opposition and even if there’s heat and friction to the whole thing. And one then tries to find a constructive path through it.

Again, I don’t want to pretend that I’m somehow so virtuous that I always make good things out of what can be bad, but it has enabled me to look at something like AI, which to me seems like an absolutely certain major disruption in civilization, and say mainly how do we shimmer with this, rather than shake? How do we quiver rather than quake? How do we see this positively? And I think it’s easy to begin to see the positive when you think of it augmenting human effort rather than replacing human effort. It would be naïve to think that there will not be redundancy of some things, as there have been with every innovation for millennia. There will also be new jobs created by this innovation.

But I think the thing that most heartens me about AI mixing amicably with publishing is the opportunity it gives for a few positive things. One, for example, is I talk about a book called Momotar?: The Peach Boy, which is a Japanese tale that I found when my children were smaller, and I brought back to them, and it was fortunately translated into English. I was able to immerse them a little bit in Japanese values and to listen to the way they thought about it. I think the spread of different cultural artifacts will be enabled and assisted by AI. So I look forward to reading a Mongolian criminal novel and also to immersing myself into why fishing in Tennessee is different from the way it is in the Atlantic or something – all this stuff is enabled by an AI making things more accessible.

I think it’ll also largely free us, as we come to accept it, from rote activities, routine activities, anything that’s predictable. Things that are predictable are really good for AI, because that’s all it is. It’s a prediction machine. So where we can say it’s really predictable what we have to do here, why wouldn’t we put AI to work there, making us essentially freer to be random, to be creative, to be unpredictable, which I think will become the definition of human essence – that we can’t be entirely predicted, and that’s the beauty of the human existence.

KENNEALLY: You founded and led what became the world’s largest qualitative market research company, Sadek Wynberg Research, which WPP, the global communications mega-firm, acquired in 2003. Nadim Sadek, what does market research on publishing in 2023 tell you are the best ways publishers can put AI to work?

SADEK: The thing that I learned about market research, and it was my particular specialty in qualitative market research, was to understand the motivations of people. What causes them to bond with other things? Often perhaps trivially, it was with brands of things, but it could also be with concepts and movements and philosophies.

So I have a sensitivity from that training to what draws people from where they are to something else. I’ve learned that that mainly comes from a sense of shared values and where you think that there’ll be a positive exchange. So if I give you my time, my energy, my intellect, perhaps even my advocacy and my money, which is kind of what’s involved in buying a book, I would hope to get something back in return perhaps that even exceeds my investment in connecting.

The thing that tends to drive the greatest sense of reward is where something is psychologically satisfying, where you finish reading whatever the matter is and say to yourself, that was a really good read, or that enlarged my point of view on it, or that was very invigorating, or that just made me think a bit differently. Whatever it does to you – and that’s the wonderful thing about literature, it can do 100 different things to you – whatever it is, that is a positive exchange. And if we with AI – and I believe we can – can identify the DNA of a book, can say this is the beauty that resides within these pages, and it should turn you on, because you like this sort of stuff, and make that match, then the prospect of having wholly more fulfilled readers who don’t after the first chapter say, God, do I have to keep going through all of this, or shall I just abandon this $15 investment of mine, and that’s another book I shouldn’t have bought? I would love to get past that, where we’re really serving up psychologically matched, sort of guaranteed fulfilling matches between authors and their work and readers who are hungry for it.

KENNEALLY: Indeed, you cite a 2023 survey showing that 70% of readers want to be matched with and buy books that are suited to their taste in a more effective way than currently done. So briefly tell us about the technology behind Shimmr.AI and how it will address this.

SADEK: Yeah, it was really interesting doing that survey. We did it in the US and the UK in January of this year, I think, 2023. And essentially what it showed is that there’s some really heavy readers. There’s a lot of light readers who have carried guilt about being lighter than they want to be. And then there are the no way am I ever reading anything again. I just look at my phone. We don’t address those guys at all at the moment, nor do we address the really heavy readers who know what they’re doing and can navigate the world of publications for themselves.

But the ones who are the majority, who told us that they would read up to 12 books more a year if they could be matched with something they knew they would really enjoy – that’s a book a month. That’s a big change to publishing economics. They essentially are suited by our technology.

Your question was what is the technology? Here it is. I think it’s easiest to think of our technology as three machines sitting in a room. The first machine we’ll call the analyzer. It’s essentially a sort of tuned version of a large language model, and we get, with the whole permission of the author and the publisher, their EPUB, and we give it to the analyzer, the first machine. Nothing that we put into it goes into the LLM’s training set. It’s hermetically sealed, and it just does the job that we ask it to do.

It extracts what we call book DNA. Book DNA is constituted by two different components, really. One is structural analysis – what’s the genre? What’s the length? What’s going on in it? And the other one is a psychological profile. So it might be a book set in Norway about mackerel fishing, but it’s also about melancholy, depression, and cruelty. That’s the DNA of this book. And the analyzer pops up and says I’ve got a slightly odd one here. It’s mackerel fishing in Norway – melancholy, cruelty, and depression. Could you make some ads for it?

And it says that to the generator, which is the second machine. The generator says, cool, yeah, I can do that. It does its thing, which is essentially using diffusion models, where it takes these very specialized prompts that we’re creating with that DNA and it creates to start with about five different assets, which are a mixture of images and copy, and then it’s kind of constituted in the right format to go into the Google environment or Amazon or wherever it is that we’re going to place these advertisements.

And it pops up really rather proud of itself and it says, I’ve made these five great ads for this slightly strange book, and it talks to the deployer, the third machine, and says, could you get them out there? The deployer says, sure, that’s what I do. It has been talking to the analyzer and says, what sort of people are going to like this? People into mackerel fishing in Norway who are miserable and depressed and into a bit of cruelty. OK, well, I know where to find those, because we do contextual targeting, where we look at not only high interest in things, but high intent to engage with them. So if you’ve been looking at a Norwegian tourism website, you’ve been looking at mackerel fishing, and you’ve been looking at I think that I might be depressed. What do I do about it? Shall I go fishing in Norway? This is probably the book to serve to you at this point in your life. And when we do, you will say, God, what a perfect match to the way I’m feeling these days.

So the technology is sensitive to how we are as people, and it’s faithful to the content of the book, and it says, I will manifest the wonders that are contained within the book that so far, by the way, have only ever been shown by the book cover, which tries to do an amazing job of talking of the thousand pages contained within its covers. We say we love the covers – not touching them, not competing with them in any way. We just go in and read the pages, and we say this is what it feels like in here, in these pages, serve it up as a series of advertising executions which go on for the whole year, and we place them with people who we know have an interest and a propensity to engage with those psychological values, the net effect of which is to surface titles that have lain dormant to audiences that you’re likely to turn on, because they want this stuff.

That produces for the reader a more fulfilling read, for the author greater fame and reach, and for the publisher durably higher revenues per title. If there is a bit of nobility to this, that’s where it comes from. There’s nobody really being harmed here. And it’s all AI doing it. That’s why I talk about shimmering, not shaking. There’s a positivity to the way that we can embrace AI in publishing.

KENNEALLY: Indeed, if publishing is going to shimmer and not shake, as you say, with AI, by AI, how much time does it have to make friends with this new technology? Is there a cutoff moment when this moves from opportunity to apocalypse?

SADEK: Well, let’s be modest and humble. We don’t know where AI may end up. And there are many very smart people who are doomsayers, who say it’ll be terrible and the end of civilization. So I cannot wholly dismiss that. It’s not what I subscribe to. It’s not what I believe in. But that school of thought exists. So maybe there’ll be an apocalypse. I cannot see it coming.

I see an inexorable adoption of the technology. It’ll come in, as I suggested, with rote, routine things. It’ll increase warehouse efficiency. It’ll make rights management simpler than it is. It’ll make editing probably a bit more streamlined than it is. It should be able to identify cultural trends so that commissioning editors can actually know the books that are going to be more in the zeitgeist of the world. All of that stuff is just really good to have. Why would we fight against it?

I think where we might end up – and this is just to go into a lot of bit of almost amateur neuroscience for a second – we have two systems as human beings. One is what’s generally called system one – Daniel Kahneman talked about it a lot – which is our instinctive, implicit, gut way of responding to the world. And the second one is system two, which is explicit and rational and cognitive and very weighed up. I think what AI as our companion in life as civilization goes on will do is to more and more say I can do the system two stuff probably faster and often a bit better than you. So I’ll land the airplane. I’ll drive the cars. I’ll do the traffic management. I’ll make sure the sewage goes the right place. All that sort of stuff will be done in a much more kind of rigorous, logical fashion by AI.

What does that leave us to do? The stuff we really like doing – chatting, writing, imagining, creating, making love, trying not to make war – you know, doing all the good things that we enjoy doing. And I think that’s really good. I love the idea that our amazing brains have created this alternative brain that we can train to be a positive companion. For sure, bad actors could make it mean and nasty, and it no doubt will go that way in some cases. But by and large, the positive embrace of AI and the way it liberates us to be more creative, to be more genius-like, just fascinates me and pleases me.

KENNEALLY: Nadim Sadek, author of Shimmer, don’t Shake, just out from Mensch Publishing, and founder of Shimmr.AI, thank you for joining me on the program.

SADEK: It’s a complete pleasure. Thank you for listening to me and asking such great questions.

KENNEALLY: That’s all for now. Our producer is Jeremy Brieske of Burst Marketing. You can subscribe to the program wherever you go for podcasts, and please do follow us on Twitter and on Facebook. You can also find Velocity of Content on YouTube as part of the CCC channel. I’m Christopher Kenneally. Thanks for listening.