“Sam Altman Would Hate It.” My Book Obsolete is Available for Preorder (and Ships in June)!

On the effort to render you obsolete and the only surefire way to stop it

Sorry I’ve been away. This is why.

After nearly three years, I (finally!) finished my book Obsolete: The AI Industry’s Trillion-Dollar Race to Replace You—and How to Stop It.

If you want to help the book, the most useful thing you can do right now is preorder and forward this post to friends. preordering through my publisher gets you the book in June (months before wide release Sept. 15), supports an indie press doing important work, and shows there’s an audience, which matters as we try to get it in front of more people. It also gets you 15% off! [Edit note: to accommodate production realities following strong preorder demand, we pushed the wide release to Sept. 15]

And if you’d like to share it on socials: Twitter, Bluesky, LinkedIn, Threads.

If you have a podcast, newsletter, publication, or other platform where you might cover the book, email press@obsoletebook.org and include a bit about your work to request a review copy. If you’d like to order copies for a class, group, or organization, email bulk@obsoletebook.org for tiered discounts.

Writing this has involved sacrifices. Mostly for everyone around me. My new housemate Tara, for instance, feels betrayed. She was told, “Garrison is a lot of fun and throws great parties.” Lies, she said. The only Garrison she knows has spent 12 hours of basically every day obliterating the vibes on the 3rd floor of our house with a hideous SAD lamp. And when he wasn’t working, he was yapping about what he learned that day, like some choice quotes about how unthinkable the Soviets and Americans actually found nuclear war.

I’m sorry Tara. I swear I’ll be back (right after my book tour).

In a nutshell, the book is about the largest project in history: the attempt to render you obsolete. The thing I wrestled with the most was: what the hell are we supposed to do about it? When you lay out all the potential horrors presented by the advent of machines that can fully replace human labor, there is really only one answer that adequately avoids them: stop. This, however, is anathema to the people that need to be convinced, so I, like many others, tried to find proposals that would help improve our chances that things go well that wouldn’t get me written off as naive (e.g. “Build a pause button”).

But in an effort to be taken seriously by Very Serious People, I wasn’t actually being serious myself. And the people in charge aren’t going to do what’s needed unless enough of us demand it of them.

Of course, thinking about what I refer to in the book as the “Obsoleting Project” is a tricky business. The best reason to believe it might succeed is also the best reason to think our obsolescence is inevitable: the unprecedented scale of the resources pouring into it. Could it really be stopped? The first person I needed to convince was myself. Researching the book, especially how the Nuclear Freeze movement got Reagan to about-face on arms control (covered in the final chapter), paired with watching concern about AI explode into the public consciousness, convinced me that stopping is ambitious, but eminently achievable.

Obsolete draws on hundreds of conversations with leading AI researchers, advocates, and industry insiders (including Geoffrey Hinton, Yoshua Bengio, Daron Acemoğlu, Eliezer Yudkowsky, Shane Legg, Helen Toner, Deborah Raji, Stuart Russell, Ajeya Cotra, Daniel Kokotajlo, and Dan Hendrycks).

With it, I took on a dual mission: get anyone up to speed on the state of AI and its risks (especially skeptics and lefties) AND write something insiders will enjoy and learn from. An unbiased panel of early reviewers (i.e. people I know) think it succeeds. For instance, here’s a review from a skeptical friend:

It felt like I was being drawn into this profound vision I had managed to keep out of my head for years… I feel this deep, visceral distrust of AI companies, and chatbots, for reasons that you were able to lay out in various places. I think I came to this like a very non-SF, non-tech, ‘normie’ who is reacting with their gut to seeing the proliferation of ai... But I also had never directly faced the story of AI, how the tech worked, and whether my skepticism was in fact, ill-founded. This hit that, and I thought it did a great job balancing the lines without getting too drawn into the hype.

And here’s one from Cambridge existential risk scholar and Goliath’s Curse author, Luke Kemp: “The best book I’ve read on AI. It’s supremely timely, it takes no prisoners, and Sam Altman would hate it.”

The book covers core concepts from AI safety, but also where they fall short, while proposing entirely new ways to think about it. Here are some of the ideas I’m most excited about:

• The Obsoleting Project (more on this in the excerpt below)

• AI reform - The technology’s biggest doomsayers and its most fervent critics are locked in an argument over whether today’s AI is a wildly useful tool or a generational scam. I contend that deep learning is incredible, but it’s being pointed at the wrong things by the wrong people for the wrong reasons. What’s needed is a new perspective on the technology, what I call AI reform, that breaks from AI safety’s narrow focus on technical solutions and the skeptics who reflexively dismiss the technology’s potential.

• The Alignment Polycrisis - The classic alignment problem just covers technical alignment—getting machines to do what you want—and (sometimes) normative alignment—wanting the right things. But this framing misses huge parts of the challenge. To widen our aperture, I introduce the Alignment Polycrisis. Technical and normative alignment are just the first two layers of the Polycrisis. The third layer is economic: we should expect unregulated competitors to do whatever amount of alignment helps them compete best, as they sprint to render humanity obsolete. The fourth is geopolitical: the specter of foreign adversaries, often cynically invoked, drives a self-fulfilling race to the bottom. Finally, a “solution” to the alignment problem is neither necessary nor sufficient for solving the problems posed by the Obsoleting Project.

• The Class of 2034 - Today’s AI revolution has been powered by deep learning, which really started working in 2012. If the deep learning era were a person, it would only be fourteen. At six, it could barely string a sentence together. By twelve, it was writing passable code. Now it’s solving problems that stumped the “master of algorithms” and is doing year-long engineering projects in an hour. In 2034, when kids born in 2012 will be graduating college, what jobs will be left for them to do?

Stay tuned for updates on excerpts, tour dates, and more.

Table of contents

Part 1 - What You’ve Heard

Prologue - The Obsoleting Project

Chapter 1 - What is Artificial Intelligence?

Chapter 2 - Is AI a Bubble?

Chapter 3 - Will the Robots Take Our Jobs?

Chapter 4 - “Just Unplug It”

Chapter 5 - “Increasingly Smart Sociopaths”

Chapter 6 - Why Build the Doom Machine?

Chapter 7 - Is AI Impossible?

Chapter 8 - The “Doomers” Feel Misunderstood

Chapter 9 - Is AI Inevitable?

Interlude - Chapter 10 - Moving Fast and Breaking Minds

Part 2 - How It Could Go

Chapter 11 - The Wrong Kind of AI

Chapter 12 - The Problems with the Alignment Problem

Chapter 13 - How Fast Could It Happen?

Chapter 14 - The Best-Case Scenario

Chapter 15 - How We Could Lose

Part 3 - What to Do

Chapter 16 - Stopping the Obsoleting Project

Chapter 17 - Actually Democratic AI

Chapter 18 - What Should the Rest of Us Do About It?

Excerpt - The Obsoleting Project

A tiny group of people, backed by the best-resourced organizations in the world, is attempting to render you obsolete. Not every participant in the project even believes this is possible, but the effort’s vanguard has defined itself through its faith in the transformative potential of machine intelligence and its willingness to forge ahead despite the grave risks it so routinely acknowledges.

The industry’s ultimate goal and the most controversial term in this space—artificial general intelligence—conjures up images of a new type of mind that might think like our own. But it would not. And “AGI” keeps the focus on what it might be, when the attention should really be on what it might do. It would be better to understand what companies like OpenAI, Anthropic, and Google are trying to build not as a new type of brain, but rather a new type of machine, one that doesn’t make products or services, but produces labor itself.

Why? Humans—and human-derived labor—are messy. Humanity has always been constrained because humans are the constraint. It takes a long time to make more of us, to educate us, to feed and shelter us. We can only do so much. For centuries, automation has allowed owners to substitute some capital for labor—the tractor for the farmhand, the power loom for the weaver—but only to a point.

Now, machine intelligence promises the potential to cut out that last constraint: us. This obliterates the one fundamental limit and opens up new worlds of possibility. Many of them, as we will see, are terrifying.

The leaders of what, in this book, I’ll call the Obsoleting Project may contest this framing, offering some bromides about how AI will augment human work or how we’ll simply find new jobs to do. But their stated goals are to create systems that so clearly surpass our abilities that we will have little hope of competing. OpenAI makes this explicit, defining AGI as “a highly autonomous system that outperforms humans at most economically valuable work.”

Many are skeptical that the Obsoleting Project will succeed. Perhaps you are too, and I don’t blame you. Often, seemingly equally credentialed people are making opposite claims about the present and future of the technology. But the Project’s scale has no real precedent. What we think of as big—Gilded Age monopolies, the Manhattan Project, the Apollo program—doesn’t even come close to capturing its size. It’s worth looking carefully at whether it has a chance of succeeding.

As with all new technologies, the Obsoleting Machine, if it works as intended, will create winners and losers. If you’d like to double down on today’s biggest winners—the ones developing it, the capital-owners, and the bosses, along with the nations and fortunes large enough to buy a seat at the table—then bring on the acceleration. The Obsoleting Project is, essentially, an inequality engine, aspirationally turning unfathomable amounts of capital into other people’s unemployment, all to maximize value for shareholders. The crowning irony: the Project was built on the backs of humanity’s collective labor, appropriated wholesale—our shared inheritance plundered to create private fortunes.

Understanding these dynamics—who wins, who loses, and why—is essential to grasping what’s actually happening right now, and what might happen next. Other books have been written about machines that might one day think. This is the first book about machines that might one day produce labor.