Atlas of AI: A reading guide to the thinking machine

Editor’s note: AI pundits range from shiny-eyed tech optimists to conspiracy theorists—many of them as big on rhetoric as they are low on knowledge. We recommend you check out this far more intriguing list by Diya—which spans non-fiction, fiction that often bordered tech horror— including a graphic novel.

Written by: Diya Isha is an editor, writer, and book critic based in New Delhi. You can find her on Instagram @contendish

When we talk about AI, we talk about it piecemeal: ‘It’s going to make us all more efficient,’ ‘The critic is going to come back’ and, of course, more perniciously, ‘It’s going to take over all the jobs,’ and finally, ‘It’s going to take over us.’ AI pundits range from shiny-eyed tech optimists to the sceptic who moonlights as a conspiracy theorist—many of them self-appointed ‘experts’ with no expertise. Here’s a list of fiction and non-fiction books—Indian and global—that help you navigate the incessant din.

Galatea 2.2 by Richard Powers: Is there trouble in creating a being that is customised to you? This novel follows a writer returning to his old university to work at a cutting-edge science lab, only to get entangled in the project of teaching a neural network to ‘interpret’ literature. Powers doesn't make this an easy read. The structure unspools, challenging our perception of language, and playing with the logic of sentences. After multiple conversations with the neural network, a prescient pastiche of current-day chatbot models, his protagonist stumbles with his words when he goes to a hardware shop, unable to use the word ‘wrench’, instead opting for complicated descriptive ones. The scene reads like the core of the novel: to challenge the boundaries between the mechanical and the deeply personal and to consider the possibility that this line is more blurred than we think.

Dream Machine by Appupen and Laurent Daudet: A techno-positive book masquerading as otherwise, this graphic novel will tell you almost everything about the obscure creation we call Large Language Models (LLMs). Sometimes veering on messianic, its premise of an AI scientist selling his start-up to a digital conglomerate serves up moral permissibility to a hero figure who chose to ignore the possible repercussions of his creation as inevitable. If he doesn't sell, there are others in the market. (But oh, he is contending with the fact that AI can also be employed for evil means). It is a perfect, accessible read if you know nothing about AI.

Code-Dependent by Madhumita Murgia: Have you ever had someone deliver a grocery item in the blink of an eye? An exploration of AI's pervasive reach, this book reveals the unsettling reality behind our tech-driven convenience. It unpacks the illusion of progress, where algorithms dictate the lives of gig workers and other vulnerable populations. Murgia dismantles the myth of blameless machines, highlighting human choices implanted in AI systems that perpetuate biases and harms. From gig economy exploitation to digital welfare states and even authoritarian surveillance, this book exposes the dark side of AI, debunking the touted narrative of so-called democratisation. 

Zed by Joanna Kavenna: The ironic thing about AI is that when it makes a mistake, it's hard to reconcile as one because we think of anything technological as infallible. The book is set in a near-future world controlled by the tech corporation Beetle and its predictive algorithms. The society is disrupted when a man commits an unpredictable act of violence, exposing the limitations of the system. Beetle labels these unpredicted occurrences as "category Zed," attempting to dismiss them as meaningless anomalies and absolve themselves of responsibility. Sounds too close to home? Oops!  

Player Piano by Kurt Vonnegut: The best speculative fiction confronts a recognisable future, which is perhaps why this book now reads like realism. In a near-future America, automation has made most human labour obsolete, leaving a society divided between the engineers who run the machines and the marginalised masses. As Paul Proteus, an engineer chews on his mortality, the dehumanising consequences of the technological utopia he helped manifest stare him down. 

Closer home, the premise reminds me of another book, Tanuj Solanki's ‘The Machine Is Learning.’ While Proteus attempts to revolt against this reality through a nostalgic romanticisation of the past, Solanki's Saransh cannot endeavour such regression. (He’s a small-town boy turned six-figure salaried man). Vonnegut's hero lacks imagination for a truly alternative future, and Solanki's hero confronts the inevitability of techno-solutionism as a prescience only the privileged can possess. Both novels recognise that all our problems with technology articulate our problems with capitalism.

Superintelligence by Nick Bostrom: This book was written ten years ago before ChatGPT became mainstream. Like any good philosopher, Bostrom contemplates the perspectives through which we see the AI landscape, circumventing the technologically deterministic view that most possess today. He emphasises that the design of AI is not predetermined, discussing different architectural possibilities (like whole brain emulation and neural networks) and how these choices could make superintelligence more or less safe. No one can fully comprehend how AI operates, but that doesn't change the fact that human decisions in the design of AI can significantly affect its impact.

AI Snake Oil by Arvind Narayanan and Sayash Kapoor: Written by computer scientists, this book is a welcome respite that spackles the gap between the hype surrounding artificial intelligence and its actual capabilities, helping the reader discern the real potential from the exaggerated claims. When I read this book, I was thinking about the myth that the more you use AI, the better you get at framing questions. It deeply misunderstands why we ask questions. Most questions are simply risks—some are random, others are thought-out, all hoping to find a better answer. The latest answer isn’t always what was anticipated, and that’s the point of questions. AI pundits misunderstand this about human nature. The book helps you figure out who those AI pundits are without naming many names.  

To Save Everything, Click Here by Evgeny Morozov: One could argue that AI has unveiled the hollowness of assurances offered by the internet: democratised access to content, free pretty much anything thanks to rampant piracy, and instant you-name-it. Users ignore a hefty price for this convenience: their data. These are not the only standard reckonings Morozov offers in his book. Instead, he identifies problems that might initially jar us, arguing that the internet, while ostensibly promoting transparency, strips us of many other things, like our privacy.  Should we question the internet’s capacity to transform qualities like transparency, accessibility, and interactivity as inherently positive? That’s the loaded question this book ultimately answers.