AI and I: Looking into the Mirror of Artificial Intelligence with Mike Barlow

What do you think about when you think about AI?

Academic integrity? Workplace efficiencies? Robot apocalypse?

Mike Barlow thinks about Hamlet, Dante, The Twilight Zone, The X-Files, Edgar Allan Poe, Gabriel Garcia Marquez. He thinks about mirrors, meteorology, and Taoism.

Barlow is the writer of a new book, The Thinking Mirror: Entangled Possibilities, the latest book released by The University of Bridgeport Press. There are a lot of books out there about AI, but none of them resemble The Thinking Mirror. Reading The Thinking Mirror is like tumbling into Wonderland or jacking in to The Matrix. Reality is constantly distorting itself. Truths you had taken for granted lose their coherence and take on surprising new shapes and relationships. Inner/outer, nature/culture, animate/inanimate, self/other - all of our familiar categories - begin to curl in on one another. And the freakiest permutation of all is that we never left Kansas. This is our world, our reality, in the Age of AI.

Wild Knights caught up with Mike Barlow to ask him a few questions about The Thinking Mirror.

WK: You’ve been writing about the confluence of culture and technology for many years, so it’s not surprising that you have turned your attention to AI.  How has your previous work writing about data science, social media, and smart cities informed the way you think about AI?

MB: Yes, I’ve been writing about AI for more than a decade, mostly because modern AI is really a combination of technologies and techniques that have been converging since around 2010. First, you needed powerful computers and high-speed data networks. Next, you needed huge amounts of structured and unstructured data, which was often called “big data.” Then, you needed an overarching philosophical framework, which technology writers like me loosely referred to as “data science.”

But you also needed imaginative foundational methodologies such as machine learning (ML), supervised learning and unsupervised learning. And then you needed new core architectures and computational models such as artificial neural networks and deep learning. And then, you needed specialized techniques like convolutional neural networks, backpropagation and transformer architectures, which were the “breakthrough” that led to large language models (LLMs) and generative AI, which is what we’re all worrying about today.

I guess this is all a long way of saying that I’ve been writing and thinking about this stuff for 15 years, and most of it still strikes me as astonishing. It also led me to wonder if modern AI was invented or discovered, since it always seemed to me that many of the components of modern AI came together more by accident than by design. Even after all of these years, I still think it’s freaky and disturbing.

WK: There are a lot of books out there about AI, but The Thinking Mirror stands out to me for its emphasis on the strangeness of AI.  Much of the book is dedicated to finding metaphors and analogies for illustrating the dream-like uncanniness of a technology that seems like it's rapidly becoming a taken-for-granted feature of our daily environment.  Why do you think it’s important to foreground the strangeness of AI?

MB: The old forms of AI, the kinds suggested by fictional characters such as the HAL 9000 in 2001: A Space Odyssey or the M5 computer in Star Trek, were based on the idea that if you built a big enough computer and then programmed it with enough information, it would become “smart.” That idea works in fiction, but not in real life.

It turns out that the “secret ingredient” to making AI work isn’t telling an AI system what to do. The “secret ingredient” is allowing the AI system to ingest a bunch of data and then letting the system teach itself. Hence the term “machine learning.”

The simple fact that modern AI systems learn from their experience makes them very different from any of our previous technologies. Turbines don’t teach themselves how to spin. Jackhammers don’t teach themselves how to jackhammer. Blenders don’t teach themselves how to blend. Prior to the rise of modern AI, all of our technologies required human beings to operate them. We turned them on, we used them to do what we wanted them to do, and then we turned them off.

At this point, there’s really no way to “turn off” AI without shutting down the internet and dismantling the power grid. To me, this seems like a strange predicament. How did we suddenly get into a situation where we have a machine that we can’t turn off?

So, yes, I find the whole situation to be very strange, and that’s why the book hovers on a cloud of strangeness. Anyone who isn’t a little bit afraid of AI hasn’t been paying attention. As I write in the book, it feels like we’re wandering in a hall of mirrors, looking for an exit sign.  

WK: Your discussion of AI is informed by creative engagement with literary writers including Poe, Spinoza, Shakespeare, and Dante.  How would you characterize the value of literature as a tool for understanding contemporary reality? 

MB: I grew up reading books. In college, I majored in English Literature. I spent most of my professional life as a writer and editor. So, I guess you could say I value words and I value writing. All my life, I’ve looked at the world through the lens of writing. I’ve admired writers who weren’t afraid to look at the world around them and write down what they saw. Whenever I’m feeling lost or unsure of what’s going on – which is pretty much all the time, nowadays – I turn to writers for advice and comfort, even if they’ve been dead for a couple of hundred years.

WK: What role, if any, did AI play in your own writing process as you developed this book?

MB: I used ChatGPT to check my work for grammar, typos and spelling errors. Early on, I thought I could “prompt” ChatGPT into writing sections of text for me, but its writing style is so cramped and formal that I gave up on that idea pretty quickly. I also tried using ChatGPT as an editor, but it always over-edited my prose. It would delete anything that I wrote that was offbeat, quirky, or innovative. It was helpful in later stages when I had a lot of text written and couldn’t exactly remember if I’d already fully covered a topic or just mentioned it. I wish I could say it was more helpful with the book’s index, but I pretty much had to do the index by hand.

At this point, I use ChatGPT mostly for brainstorming off-the-wall ideas that none of my friends or relatives have the time or patience to talk about with me. It’s also useful when I have a weird dream or wake up with a crazy idea at 2:30 in the morning and need to capture it quickly before it vanishes from my memory. Yes, I keep a pen and paper by my pillow, but I can never read my handwriting the next morning.  

WK: What effect do you want The Thinking Mirror to have on readers?  How do you hope it informs the way they relate to and interact with AI in their work and in their lives?

MB: I hope readers find the book interesting and amusing. I hope that it reminds them that AI is more than just new technology. I hope it also reminds them to make sure that when they use an LLM, they need to be skeptical about the output. Just because an LLM’s response seems to make sense doesn’t mean its response is correct or appropriate. Don’t confuse fluency for intelligence. In other words, AI is fast, but only you, the human being in the loop, can decide if what it’s telling you makes sense. Don’t surrender your agency.

Let’s not kid ourselves. We’re experiencing the beginning of a new era in human development. That’s not necessarily a good thing. Like most people, I’m not a fan of overly dramatic transformations. I’d prefer if things changed more slowly and gradually. AI is making me think that maybe the Middle Ages weren’t so bad.  By the way, the book’s original title was “An English Major’s Perspective on Artificial Intelligence,” but my wife told me to think of a new title. 

Mike Barlow