The Reverse Centaur’s Guide to Life After AI; How to Think About Artificial Intelligence — Before it’s Too Late, Cory Doctorow, 2026
A reverse centaur is a machine that uses a human conscripted to serve as a biological appendage for a machine working at an inhuman pace.
The fact that there’s a low probability that an AI will be able to do your job doesn’t change the fact that there’s a high probability that an AI salesman will convince your boss to fire you and replace you with an AI that can’t do your job.
Never forget that you aren’t the target for AI hype — investors are…If you drive 101 of 280 past (SFO) or San Jose (Airport) , you’ll pass gigantic electronic billboards, pumping out ten of thousands of ANSI lumens that glow even at high noon, seen by thousands of commuters but there to pitch only a couple dozen VCs and executives at major firms.
If you want to puncture the AI bubble, you should train your fire on the applications that are used to justify the massive investment in data centers and training.
The workers who are dead center in the crosshairs of AI bosses are programmers. Google, Amazon, Microsoft, Apple — over and over, we hear announcements from tech bosses about how many of their coders they plan to fire once the AI works, or (even more ominously) how many coders they’ve already fired because AI works so well.
The reason tech workers are able to command all these on-the-job goodies is down to an accident of history: when computers were absorbed into every kind of industrial and personal activity, the supply of trained coders was nowhere near high enough to meet the demand for their obscure, hard-to-master skills.
This meant that coders could demand all kinds of concessions from their bosses because there were always high-paying jobs with gobs of perks going for anyone who knows how to turn out reliable code on deadline, and bosses could afford to meet those demands and still turn gigantic profits.
In 2018, Google workers by the tens of thousands, walked off the job, kicking off a series of confrontations that forced the company to abandon a censored search engine for the Chinese market, a $10 Billion military project…The exec in charge of the military contract resigned…In the space of just a few months Google declared its first dividend, fired twelve thousand workers (including many of its most senior–and thus most mouthy–technical staff), and declared a $70 billion stock buyback, which would have paid those workers wages for the next twenty-seven years.
One Google engineer relates his experience with AI in the workplace: “I have been a software engineer at Google for several years. With the introduction of generative AI-based coding assistance tools, we are already seeing a decline in open-source code quality (defined as ‘code churn’ how often a piece of code is written only to be deleted or fixed within a short time). I am also starting to see a downward trend of (a) new engineer’s readiness in doing this work, (b) engineers willingness to learn new things, and (c) engineers effort to put in serious thoughts in the work.”
For AI companies to make back the hundreds of billions, their investors have entrusted them with, they will have to displace a hell of a lot of high-waged labor. That’s displace, not augment. AI companies are selling the replacement of workers with chatbots, but chatbots just can’t do workers’ jobs. To sell hundreds of billions of dollars worth of AI, you need a killer demo.
During the drafting of this book (2026), an MIT study found that 95 percent of commercial AI deployments fail, with “no measurable impact on profit.” The news sparked a panicked sell-off of AI related stocks, though whether this is the pin that pricks the bubble remains to be seen.
Its essential that we never stop reminding people that the current, actually existing lucrative uses for AI are terrible and should be banned.
In 2025 builder.ai (once valued at more than $1 billion) collapsed…In reality, builder.ai was a secret employment agency, farming out the work of building its customer apps to eight hundred to one thousand low waged Indian programmers. Wags said the “GPT” in ChatGPT stands for “Gujarati People Typing”.
Tech bubbles are surprisingly easy to generate, thanks to something economists call “the Byzantine premium.” That’s the extra value that investors place on an asset that they don’t understand.
Every bubble is a transfer of wealth from savers to crooks. Every bubble is bad. We shouldn’t have bubbles…Regulators should intervene to prevent bubbles in the first place. ..Some bubbles pop and leave nothing behind. These are the pure fraud bubbles.
The crypto bubble keeps getting reinflated, not least because the literal president of the United States issued his own shitcoin…But eventually the crypto bubble will burst (again) (and permanently) and when it does what will be left?
The environmental costs of the “compute” is off the charts. Even if you stipulate that the world will benefit from having some giant “advanced” AI tools, there’s no rational case for endangering the planet and the lives of millions of people to make several redundant AI tools that are functionally indistinguishable, with each consuming so much energy that they wipe a substantial share of the gains made from solarization and the broader switch to renewables.
Remember: seven giant AI companies account for 35% of the U.S. stock market. Amputating 35% of the market is going to destroy a ton of innocent bystanders, including people whose retirement savings are invested in index funds, considered the safest of all safe bets. We’re talking about a crash that will put 2008 in the shade and meet or exceed the pandemic selloff.
What’s more, that AI center is literally incinerating million dollar GPUs all the time and these have be be constantly replaced.
You can’t give a third of the S&P 500’s value over to seven money losing AI companies that energetically pass the same $100 Billion IOU around and around without creating the conditions for a prolonged, brutal global crash.
Hangzhou DeepSeek Artificial Intelligence Basic Technology Research Co., Ltd., doing business as DeepSeek, is a Chinese artificial intelligence company that develops large language models. Based in Hangzhou, Zhejiang, DeepSeek is owned and funded by High-Flyer, a Chinese hedge fund.
The release of Deepseek in 2025 sent shock waves through AI investors. Deepseek laid bare the incredible laziness of the giant U.S. AI companies, who solved all their scaling issues by throwing money at their problems rather than by applying their ingenuity to them. Deepseek’s debut sent a cold chill up the spine of every investor in a big U.S. AI company. If their $100 billion models can be bested by a model that cost a reported $6 million to create and can run on commodity hardware, what future do these top-heavy AIs have?
But they (AI companies) haven’t invented an intelligent being. They haven’t set in motion the tools to conjure up a new god or demon. They haven’t even invented a tool that can do your job for you.