Cory Doctorow on Life After AI

Good morning. It's Friday, July 31st, 2006. This is up front. I'm Brian Upwards-Tieker. On today's show, what happens after the AI bubble? We'll still have the data centers. We'll still have, as you say, some of those power plants. Maybe some of them will even be renewable. We'll still have the GPUs. We'll have lots of technical workers who know how to mobilize those things to make useful things. Corey Doctoro joins us to discuss his new book, The Reverse Centaur's Guide to Life After AI. We'll talk about technology, automation, and power at the point of production. What a difference it makes to have technology put at the disposal of workers versus having workers put at the disposal of a new technology. That's coming up next. After the news. i'm max pringle with these headlines president trump says the militant group hamas has agreed to disarm in hand over power to an interim coalition and end the gaza war but it will face many hurdles and could take a long time to achieve if at all more from simon marks The Board of Peace, which of course President Trump created primarily to try and restore peace and security to Gaza, said last night that Hamas had agreed to a detailed roadmap on the disarmament of its forces and the Israeli withdrawal from Gaza that would follow. What do they mean by a detailed roadmap? Is there a distinction between that and an historic agreement for complete disarmament? Who's going to go... in and do the disarming, who is going to then verify that the disarming has taken place. With FSN Spotlight, I'm Simon Marks. Spanish officials said some 60,000 migrants have entered Spain's North African enclave of Queda from Morocco in the past 24 hours. At least 34 people have died in the process. Spain has deployed its military after thousands breached the border with Morocco. Karen Chamas reports. Many of the deceased drowned while others were killed in a stampede at Tarajal Beach. Video footage on Thursday showed tronds of people, mainly Moroccans, walking around the breakwaters onto local roads. Salat from Tetuan in Morocco told the AP, It's a bad life in Morocco. That's why we get swim into hand. Spanish Prime Minister Pedro Sanchez is expected to visit Theota on Friday morning. A local official called the situation a serious humanitarian crisis, saying thousands of migrants, including unaccompanied children, were left sleeping in parks and sidewalks, while others roamed the streets aimlessly. I'm Karen Shamas. Major oil companies reported massive profits while fighting between Iran and the U.S.

disrupted oil shipments through the Strait of Hormuz. Prices for Brent crude soared above $100 a barrel for much of the spring. Constrained supplies meant consumers around the world paid more for fuel and confronted shortages. But higher prices resulted in elevated profits for some of the biggest publicly traded oil companies. American oil and gas giants Exxon Mobil and Chevron both more than doubled their profits in the second quarter, their position to benefit because they own refineries and extract oil and gas. Democrats in Congress have proposed taxing windfall oil profits. The death toll from the 7.1 earthquake that struck southwestern Japan earlier this week has reached at least 34 as survivors struggle in overcrowded shelters in blistering summer heat. They face shortages of water and fuel. Kumamoto's prefectural government said rescuers race to find those still missing after the quake triggered a mall explosion, toppled a factory chimney, and flattened homes. Search operations continued today, but the number of missing remains unknown. Nearly 80,000 homes in Japan or without water. firefighters in southwestern france are working to control the last hot spots of a massive wildfire the blaze has burned in area four times the size of paris forcing thousands to evacuate The fire has affected 42,000 hectares in Bordeaux's Gironde region. Government officials said today that lowered temperatures and higher humidity are helping firefighters. Europe is facing record-breaking heat, fueling wildfires across the continent. The wildfire in southwest France initially forced 224,000 people to flee. Authorities have allowed 144,000 evacuees to return this week. French Interior Minister Laurent Nunez said hotspots remain, and Ukraine is sending personnel to assist. Democratic main Senate candidate Troy Jackson is assembling a team for his run to unseat Republican incumbent Susan Collins. Jackson is stepping in at the 11th hour for the controversial oyster man Graham Platner, who stepped aside recently amid sexual assault allegations, which he's denied. Maine's Democratic Governor Janet Mills this week declined to endorse Jackson's candidacy. Several candidates recently dropped out of the race to woo 601 Democratic delegates to clinch the party nomination. However, Jackson, the former president of Maine's Senate, gained 566 votes and recently earned an endorsement from former Vice President Kamala Harris, who said she could relate to his short timeline to campaign.

main main's governor mills dropped out of the senate race in april and declined to endorse Graham Platner at the time mills has accused the incumbent republican Susan Collins of allowing president trump to abuse his authority Seattle's police chief has resigned amid criticism that the city was slow to release key information about a fatal shooting at a food festival last weekend. Mayor Katie Wilson made the announcement Thursday after hours of speculation about whether police chief Sean Barnes would step down. Some of the chief's supporters say he had rebuffed Wilson's request to resign. Wilson said she's appointing Deputy Chief Andre Sales as interim chief, Sales as a law enforcement. veteran with more than two decades of experience. The mayor said she was grateful for Barnes service, and Barnes called it an honor to have served the city. A new analysis from the British energy market think tank Ember shows that California hit a major milestone in May, generating more than half its electricity from the sun and was believed to be a global first. Observers say it's due to the state's focus on ramping up solar energy over the past five years. It's the first time a large-scale power system is known to have surpassed 50% solar for an entire month, and the second significant power record set by the state recently. For the first five months of 2026, California made more electricity from the sun than from natural gas. Sunny today in the San Francisco Bay Area, highs in the 80s, lows tonight in the 50s, and the Central Valley. A heat advisory is in effect, highs near 100, low tonight in the 60s. Those are the headlines. I'm Max Pringle on KPFA. You're listening to Upfront. I'm Brian Edwards Teakert. Very pleased to welcome back to our airwaves, Corey Docterro. He is a blogger, digital rights activist, employed by the Electronic Frontier Foundation. The author of many books, the latest of which is a cry to engage with AI, not as in evil unto itself, but in terms of who it works for and who it works on. The book is the reverse Centaur's Guide to Life After AI. Corey, Doctoro, welcome. Thank you for having me on. At this moment, the world is completely saturated with books about AI. What did you think was missing from the corpus? You know, this is a case of me making a terrible miscalculation. I was so... just tired of being dragged into discussions about AI, which frankly, I just don't think is important enough as a technology to get all the attention that it's getting. And it takes so long to explain it because it's such a hyper object that touches so many different aspects of our economic and labor discourse and so on that I thought, okay, I'm just going to write it all down and then I can stop talking about AI. And now, of course, I have to talk about AI because I have a book about it. I probably shouldn't say that, but it's the God's honest.

truth. Well, let's talk about the technology itself then to kind of set the table. What is like genuinely new and useful here? And how do you separate that from the massive amounts of hype that get into VC pitch decks? You know, the underlying computer science stuff is quite interesting and nifty. You know, fundamentally what happened was about a decade ago, some computer scientists said, hey, let's try some minor variations in the way that we run these machine learning algorithms that are pretty well understood and that, you know, we know what we put in and what we get out of them. And instead of getting a small improvement from making this small change in the way that they acted on these algorithms, they got a much bigger one, a big, big, fat, hairy improvement. That was an exciting piece of low-hanging fruit. And it turned out to be part of an orchard of low-hanging fruit, that this was a very scalable circumstance where you could keep doing the variation harder and harder, scaling it up, and you would get improvements over and over again. And here we are. We've got a bunch of improvements. in this very powerful statistical inference technique that can do things that we're all familiar with. It can generate images and edit images. It can generate text and it can converse with you in a way that can be quite convincing. It can write code. It can do a lot of things. The problem is that it's not fit for purpose for being an unsupervised autonomous, what they sometimes call an agent that just does your job. What it's good for is for people who are skilled at some task looking at the panoply of ways that it can automate aspects of those tasks, thinking about which ones they personally feel comfortable and trusting to a machine, and then allowing the machine to do those parts of it while closely supervising it to make sure that it's not making mistakes. And for it to fit that niche, the automation has to be adopted by labor, and that automation adoption has to be driven by workers. And that's not where we've landed. We have raised now over $1.4 trillion for AI on $50 billion revenue. And to turn that $50 billion into anything like a profit is going to require firing a bunch of workers and replacing them with chatbots. and taking the workers who survived the job apocalypse and putting them in charge of marking the AI's homework at superhuman speed for durations that exceed all human capacity. And then I think ultimately just taking the blame when the machine goes wrong and hurts people or kills people or wrecks things. Your argument is not that the AI technology we have is capable of replacing all those workers. It is that it is an existential necessity. for AI companies to claim that they'll be able to do that. Yeah, and also that bosses are suckers for that pitch. You know, I don't know whether the people who decided that we should spend $1.4 trillion on this technology really believe that this can do your job.

Or whether they believe, you know, your boss is really gullible when someone shows up and says, hey, you know, those workers that you have those ego shattering confrontations with because you give them your idea and they say, sorry, that's illegal or impossible or, you know, we'll go to prison or it'll maim and kill a bunch of people. And, you know, how about you just replace that with an algorithm that says, yes, boss, right away, boss and poops out the product fully formed without ever having to go through people who know what they're doing. You don't necessarily have to believe that it can do the thing to believe that... a boss can be convinced that it can do the thing. I mean, if you're investing in supplements to be hawked on Manosphere podcasts, you don't have to believe that they will make you an irresistible love god. You just have to believe that there's a lot of men out there who'll pay good money for placebos that are claimed to make you an irresistible love god. And I think the same is true of bosses who are and always have been absolute marks for claims about automation replacing workers. One of the points you make in the book is that the kind of dystopian warnings about the of the technology serve to reinforce the overinflated hype about what it's actually capable of. What looks like criticism is actually reinforcing the pitch to investors. Yeah, I'm indebted here to the STS scholar Lee Vinsal at Virginia Tech who coined this very useful term, critter hype. which is when you repeat the hype of the people pushing some kind of stock swindle, but at the end of it, you put in parentheses, and that's bad. So, you know, Vinsal started writing about this when you had a lot of people who were credulously repeating Mark Zuckerberg's claim to have reinvented himself effectively as a kind of cyber Rasputon who had built a mind control ray out of big data. And they were running around and going, look at this evil dopamine hacking wizard who runs Facebook, why he could sell anything to anyone? And then Mark Zuckerberg's salespeople went out and said, why should you pay a... 40% premium to advertise on Facebook? Well, don't take my word for it. Ask my critics. My boss is an evil dopamine hacking wizard. He is cyber-resputin and can sell anything to anyone using big data. And I think today, the version of that is... this machine is so powerful that it will take my job, not this machine is run by people who are so glib that they will convince my boss to give it my job, even though it can't do it. And you know, an extremist, the version of this is this machine is so powerful that it's going to wake up and turn into God and turn us all into paper clips. And if it's powerful enough to do that, well, then surely it can do your job on the way. There was a period when... The constituent technologies that we call AI today were at work doing a lot of things without the hype, right? Machine learning algorithms going to work on like large language corpuses were making automatic translation better, making it possible to do automatic transcription of audio and videos. What's the imperative that made the hype cycle take off that made it?

possible for these mega-scalers to corral $1.4 trillion in investment? Well, there's a funny paradox to firms that conquer their market. It's that while they're growing, And growing and growing and growing, they're market darlings. They enjoy extremely high price to earnings ratio, which is to say that for every dollar they bring in, the market values them at 20 or 30 or 50 or in the case of Tesla, sometimes hundreds of dollars. For every dollar in revenue they bring in because a share in a firm is a claim on its future earnings. And so a growing company is a company that... whose shares are worth more because it will produce more income in the future than it's producing today. So if you imagine two companies with the same earnings, but one's growing and one isn't, obviously the growing one has a higher share price. The problem is that when you conquer your market, right, if you're Google and you have a 90% search market share, you stop growing. Right? Where's your growth going to come from? I mean, you can, you know, raise a billion humans to maturity and hope that they become your customer when they reach the age of majority, which is a product called Google Classroom. But it takes like a decade for that product to turn into something that Wall Street's going to value. And in the meantime, you need to have a story that you can tell the market about how you're going to keep growing, lest the market decide that you are mature. and stage one of these mass sell-offs that we see whenever the companies post less than the robust growth that they were hoping for. And, you know, we see this with Facebook in 2022 or Nvidia. And, you know, anyone who is invested in one of these companies understands that when they stop growing, everyone else who's invested in the company is going to sell their shares. And so they are all poised to sell first, which means you get these flash crashes. Now, if you're... the management team of one of these companies, well, then your net worth is tied up and shares in the company. So that event will make you personally much poorer, but also if you want to keep the company rolling and you want to restart its growth, you will have to confront the reality that while a growing company can use its growth stock to buy other firms and hire key personnel, a mature company is expected to use dollars for those. And dollars are a thing that are hard to get. They're exogenous. You get them from, you know, lenders or you get them from customers or you get them from investors, whereas shares are endogenous. You produce them on the premises by typing zeros into spreadsheets. And so this is an incredible advantage to have, and you'll do anything you can not to lose it. We've seen this string of tech bubbles and tech stories about growth. And the early days, it was stories about how very large, arguably mature firms would find growth by cannibalizing each other.

Google Plus was going to make Google into Facebook. The pivot to video is going to make Facebook into YouTube. The problem with these is that while we can agree on how much Facebook is worth, because it's in the balance sheet every quarter, Facebook has very strong ideas about why Google won't turn into Facebook, and they make those ideas very public, and it can be hard to sell the story. And so we see this pivot to stories about imaginary markets, blockchain and metaverse and Dow's and Web3 and so on and so on. you know, you don't have to realize the vision. You just have to have a new story by the time the old story runs out, you know, to run across the river on the backs of alligators and just not lose a leg. And here we are at AI, which is the mother of all of these growth stories. It's a story about how if we just give the tech companies enough money, they will effectively create God. And then they will displace all the labor done in the economy and a share. of that company becomes a share on all the value done or created by all the jobs in the entire economy. It's kind of an astonishing feature of whatever stage of capitalism this is that you have this kind of fictitious capital, this value that only lives in the imagination of the people who are bought into the story, but you can actually spend it. You can buy chips from Nvidia by giving them stock at that incredibly inflated valuation. You can hire incredibly skilled computer scientists by giving them stock options in your incredibly inflated valuation. The story has a real material effect on how the economy gets structured. Sure. And, you know, Nvidia will actually loan you money that they get. from their stock, imaginary money, to buy their chips with, and then they'll book it as revenue. You know, we have seven companies that constitute 35% of the S&P 500. Invidia is one of them, the other six of the big hyperscalers that are all losing money to Nvidia. And in many cases, Nvidia is the one giving them the money to buy Nvidia chips. They're effectively passing around the same IOU really, really fast and pretending it's in all of their bank accounts at once. So this really is remarkable. It's also not stable. you know these arrangements these credit bubbles we have seen them before they're not a novel feature of the of the economy although the scale is novel i think that this rivals the south seas bubble or the tulip bulb mania in terms of how much of the economy it implicates and that means that there are many brittle places in this bubble and the bubble will pop I mean, there's this idea in finance, it's called Stein's Law, that anything that can't go on forever eventually stops. And so when the music stops on that IOU pass the parcel game, when some Gulf State decides that it's going to default on its obligations to finance data center build out because it has to rebuild its LNG terminals, because for reasons no one can really understand, Iran is bombing it. Perhaps it has something to do with Donald Trump as deciding we have too much oil in the world.

a thing I never thought he would ever decide, but apparently he's like out Greta and Greta, then, you know, these, you will see this cascade of failures. And I think in retrospect, we'll look at whatever it is that precipitates the crash and say, oh, obviously that was going to be the precipitating factor. But it'll just be survivor bias because there's 50 potential precipitating factors, any one of which could detonate at any moment. So the signs of a bubble, the circular finance arrangements you were just describing. the sky high valuations of these companies compared to the money they're actually bringing in. If that bubble unravels, though, the capital they've raised. It's getting spent on real things or things of economic value, not tulip bulbs. Like it's building data centers and in some cases power plants. It's purchasing very expensive graphic processing units to put in those data centers. These are things that even if their value gets marked down, they have some utility to whoever comes into them after a crash. Yeah, that's the thing I actually am hopeful about. I mean, I don't think that bubbles are good ways to produce this kind of uh residue um they're they're not just inefficient right because they're they're spending lots of money it's in stupid ways in order to uh you know sell the sell the stock swindle but they're also a system for victimizing outsiders on behalf of insiders and we just saw this with with Elon Musk space x IPO where he built the world's most efficient money furnace a company called grok a i And then he kind of hit it inside of SpaceX. And then he did sweetheart deals with the stock exchanges and the regulators so that SpaceX debuted on the stock index on the NASDAQ 100, which meant that if your retirement savings. was in a NASDAQ index fund, your fund manager was required to buy SpaceX, right? If SpaceX is 5% of the index, then 5% of the shares in your fund have to be SpaceX. Sort of by law, that's the charter of the fund. And it has to be the case at any price. Now, normally, a company that enters the stock market would be ineligible for inclusion in one of these indexes for several consecutive quarters while they were seasoned, which is to say, We made sure that they weren't, you know, part of a bubble. But Elon Musk having won this exemption was able to debut on the index. And not only that, but he was granted a further exemption from the lockout requirement that blocks insiders from selling their shares, which is another precaution because insiders know things about the company that are not known to the general public, including whether it's a money losing dog. And so generally when we expose the public markets or affirms to the... public markets and investors from the public markets to firms, we block insiders from selling for several months before they can put their shares on the market. And instead, Elon Musk was allowed to restrict the number of shares being sold by the company itself, which meant that his billionaire Silicon Valley pals and his Emirati pals and everyone else who put money into his money furnace were able to

to flog their shares at effectively infinite markups because there was no supply and there was this inelastic demand, right? Everyone who was managing an index fund had to buy the shares. They had no choice. And so we were made the exit liquidity for the insiders in this company. And I think that really shows you you know, the dynamics of the bubble at work, and it's very characteristic of these bubbles. So, you know, I really worry that every bubble is bad because every bubble, even if it leaves behind a productive residue, ends up bankrupting a bunch of ordinary savers who are just hoping, you know, not to starve to death in their dotage. And nevertheless, some bubbles do leave behind something productive. You know, Enron was a pure accounting swindle, and so it left behind nothing. WorldCom was an accounting swindle too, but in order to sell the... swindle they pretended that they had orders for lots and lots of fiber thousands and thousands of miles of it they dug up roads all over the world and put fiber underneath it and that fiber is still there Bernie Ebers the CEO who masterminded the swindle died in prison but the fiber is around today and at my home in Los Angeles I have a two gigabit symmetrical link that I get through AT&T because AT&T was able to buy some old world com fiber and light it up you know i think we can look to the bubbles of today and we look at say cryptocurrency and i think when cryptocurrency goes to zero we're not going to have anything left, you know, ugly monkey JPEGs and even ugly or Austrian economics. But when AI pops, we'll still have the data centers. We'll still have, as you say, some of those power plants, maybe some of them will even be renewable. We'll still have the GPUs. We'll have lots of technical workers who know how to mobilize those things to make useful things. One of the things that was very exciting after the dot-com crash was all the terrible bosses who had the world's worst ideas for the web. were kind of out of the picture, and their capital could be purchased for pennies on the dollar by the people who used to work for them, who could then go off and build Web 2.0, away from the madness of their foolish bosses. And I'm sure that's the case with people who work inside the AI sector. They're doubtless, many of them who have really cool ideas for ways that these things can be useful plugins and not a thing we bet the world economy on. And they're going to have access to these open source models that were released by the AI companies. often as demos or as convincers that they treat as kind of toys, but which typically are lagging the so-called frontier models just by a few months in terms of what they can do. And every time someone, you know, looks at one of these models and checks to see whether there's a way they can squeeze more performance out of them, it turns out there's lots of room for optimization. And I think the reason for that is when you're selling a stock swindle and, you know, you know that your investors don't really understand the commodity or the proposition. you need to have some kind of heuristic. You need to have some kind of rule of thumb they can follow to guess how much money they're going to get. And the most basic of these is that money in is a function of money out. So if Google put $100 billion into data centers, it's because they anticipated $2 trillion in data center returns. And that means that the more money you spend, the more money you can get normies to give you.

I sometimes liken this to the idea that if the pile of poo is big enough, it has to have a pony under it somewhere. And why would you make the pile of poo any smaller if the larger it is, the more ponies people believe you have? We're speaking to Corey Doctor O. His new book is the Reverse Centaur's Guide to Life After AI. This is the extended and intact version of an interview that we originally aired in abbreviated form a week and a half ago because of KPFA's Fund Drive. Thanks to everyone who chipped in early and generously, we were able to end that Fun Drive almost three days early. because we have to plan the finishing moment of the fun drive a little bit in advance. We didn't quite hit our target. We fell $6,000 short. But that is a highly surmountable shortfall. And if you would like to encourage KPFA's momentum towards shorter fund drives, it's not too late to give to help us make it up. The best place to donate is kpFA.org. And as it happens, Corey's book is one of the thank you gifts that we still have available on our website. www.kfa.org. And to everyone who did give and has already given as much as they can, A profound thank you. You're what keeps us going. All right. We're going to continue the conversation with Corey Dr. Rowe in the second half of our program and go into some of the second order effects of the hype cycle that the AI companies require to prop up their valuations. There's a lot of anecdotal evidence now that many companies that laid off people because of what they thought would be the labor-saving properties of artificial intelligence are now regretting those decisions. Those employees and those companies would be victims of the hype. We're also seeing massive reallocations of labor and materials in the real economy towards the construction of data centers and electric infrastructure to service them. It's part two of our conversation with Corey Doctoro next. After the news. I'm Max Pringle with these headlines. President Trump says the militant group Hamas has agreed to disarm and hand over power to an interim coalition and end the Gaza war. But it will face many hurdles and could take a long time to achieve, if at all. More from Simon Marks. The Board of Peace, which of course President Trump created primarily to try and restore peace and security to Gaza, said last night that Hamas had agreed to a detailed roadmap on the disarmament of its forces and the Israeli withdrawal from Gaza that would follow. What do they mean by a detailed roadmap? Is there a distinction between that and an historic agreement for complete disarmament? Who's going to go...

in and do the disarming, who is going to then verify that the disarming has taken place. With FSN Spotlight, I'm Simon Marks. Spanish officials said some 60,000 migrants have entered Spain's North African enclave of Queda from Morocco in the past 24 hours. At least 34 people have died in the process. Spain has deployed its military after thousands breached the border with Morocco. Karen Chamas reports. Many of the deceased drowned while others were killed in a stampede at Tarajal Beach. Video footage on Thursday showed trams of people, mainly Moroccans, walking around the breakwaters onto local roads. Salat from Tetwan in Morocco told the AP, It's a bad life in Morocco. That's why we get swim into hand. Spanish Prime Minister Pedro Sanchez is expected to visit Theota on Friday morning. A local official called the situation a serious humanitarian crisis, saying thousands of migrants, including unaccompanied children, who left sleeping in parks and sidewalks, while others roamed the streets aimlessly. I'm Karen Shamas. A new analysis from the British energy market think tank, Ember, shows that California hit a major milestone in... generating more than half its electricity from the sun and what's believed to be a global first observers say it's due to the state's focus on ramping up solar energy over the past five years It's the first time a large-scale power system is known to have surpassed 50% solar for an entire month, and the second significant power record set by the state recently. For the first five months of 2026, California made more electricity from the sun than from natural gas. Sunny today in the San Francisco Bay Area, highs in the 80s, lows tonight in the 50s, and the Central Valley. A heat advisory is in effect, highs near 100, lows tonight in the 6th. Those are the headlines. I'm Max Pringle on KPFA. Bit past 7.33 in the morning. The show is up front. My name's Brian Edwards Teigert. Thanks to the early finish of KPFA's Fund Drive, we are bringing you an interview. We originally aired a week and a half ago that we had to severely abbreviate because we needed to make room for fundraising. Our guest is Corey Doctoro. He is a prolific blogger, a digital rights activist employed by the Electronic Frontier Foundation. the author of many books of science fiction and a few of nonfiction. His most recent is the reverse Centaur's Guide to Life After AI. On the first half of our interview, he made a kind of political economy point about the imperatives facing any tech company.

that is treated as a growth stock. And that is that they have to keep telling a story about themselves to investors that describes a virtually unlimited horizon for growth. This gets their stock valued much higher relative to the money that they're actually bringing in because investors are thinking about future earnings, not present earnings. And that higher stock valuation translates into real power. Power to raise money, power to spend money, power to acquire other companies. The question is what the second order effects of the hype are. When you have trillions of dollars in market capitalization. resting on fantastic proclamations about what a technology can accomplish. What does that do to the rest of the economy? Let's go back to the conversation with Corey Doctoro. A principal concern of your book is what the hype cycle around the bubble. does to people, not because of the properties of the technology, because of the properties of the hype. So maybe you should unpack your title for us. What is a reverse centaur? Well, it won't surprise you to learn that this isn't the first time we've had a debate about automation and labor. And in the literature of automation and labor, there is a truism that when labor drives automation uptake, they do so in order to improve the quality of their outputs. And when capital drives automation uptake, it's to improve the throughput, the rate at which material is being made, often irrespective of the quality. You can see this in kind of the icons of automation, whether that's Charlie Chaplin getting sucked into the gears in modern times or Ethel and Lucy trying to get chocolates into the chocolate box at the end of the conveyor belt. The system is running at the maximum speed because it's a depreciating asset and you want to get as much value out of it as you can before it falls off your books. And so the people who are in charge of setting the tempo. run it at the absolute limit of the capacity of the slowest link in the system, which is you, the worker. And so it's always this. And this is how we get to centaurs and reverse centaurs. In automation, we call someone who is assisted by a machine, a centaur. You can think here the metaphor, the human head on the horse's body. The horse is stronger than you. It's more dexterous than you. It has more endurance than you. But it doesn't exercise judgment. You tell the horse where to go. And so whether that's you riding a bicycle or operating a spell checker or even using you know one of these interactive development environments that programmers can take advantage of that weren't around when I started programming you know those are all aspects of centaurhood and they are have the hallmark of being driven by labor uptake and labor specifying how they're used

The converse of that is a reverse centaur. It is the machine atop the human body. It is the human conscripted to serve the machine. It's the machine exercising judgment and the human following through on that. So you know, you can think here of an Amazon warehouse. It's the most automated warehouse that the world has ever seen. And it has three times the injury rate of any other warehouse in the sector. And that's not a coincidence. It's the consequence. Why would you spend seven or eight figures automating? a warehouse and then run it slower than you absolutely have to. You want to recoup that capital investment. And if you're a human being made to work at the edge of your endurance and at the edge of your dexterity and at the edge of your strength, for long periods, you will eventually make a mistake. And because the machine doesn't know who you are or how to relate to you, it's not another person. It doesn't know that you've made a mistake. And so when you have that lapse, that might be the moment you get impaled on a forklift. Is there a clear delineation? I mean, like one of your earlier examples, a spell checker, I guess makes you a centaur. But also kind of locks you into a context, right? If the spell checker is part of the word processor in Microsoft 365, it also puts you in an ecosystem where like your every keystroke is producing data that's being surveilled by a massive tech company and possibly shared with your employer for the purposes of micromanaging your activities. It is both empowering you and making you work for the machine at the same time. Yeah, I mean, this is where I think we run up against a kind of vulgar Thatcherism that is practiced by the tech sector. So Margaret Thatcher was infamous for saying there is no alternative, by which she meant stop trying to think of an alternative. If you like the benefits of the market system, then you can't. have anything that is not delivered by the market system in its rawest, least regulated form. Otherwise, the market will stop working. And by the same token, I think a lot of tech bosses want us to think there is no alternative that if you have a spell checker, your boss also has to be able to count your keystrokes and deny you a raise if you don't make enough keystrokes in an hour. It's funny because Bill Gates used to make fun of IBM because they paid their programmers by the lines of code. He used to say this is the race to build the world's heaviest air. airplane. And today, Microsoft sells and uses coding tools that count how many lines of code you write and help your boss figure out how to punish you if you aren't writing enough lines of code to satisfy their urge. So the world's heaviest airplane is all around us. And I think that it's absolutely possible to decouple the spell checker from the thing that counts your keystrokes. And the reason I know that's the case is because Office didn't do that until like a couple of years ago. It wasn't that office's capacity to spy on you came down off a mountain on two stone tablets. These were decisions taken in living memory by named individuals that had this predictable and predicted outcome.

there was nothing inevitable at them and the thing that allowed it to happen was things like the earlier decisions to dismantle labor protections and to make it much harder to unionize which meant that you can't tell your boss i'm sorry you're allowed to give me a spell checker but you're not allowed to count my keystrokes and you know we can see where that's worked you know look at the writer's strike in hollywood they won the right to use AI however they wanted whenever they wanted but not but their bosses couldn't make them use it and their bosses couldn't pay them any less if they used it And so, you know, that's a very centaurish arrangement. If you want to have a chatbot go through all the scripts and the series you write on and then allow you to query it by saying, hey, what episode was it where such and such happened? And to have that come back up in the writer's room along with time code and the written script and the shot script and, you know, a link to the video in your proprietary CMS so you can watch that scene again as you're in the writer's room. That's your business. But if your boss wants to fire three quarters, of the writers room and put the remainder in charge of marking the AI's homework. That's just not on the table. And the reason they won that right was because they're unionized and not just unionized. They're super unionized. They are among the only workers in America who are exempted. when the Taft-Hartley Act was passed in 1947 from a ban on what's called sectoral bargaining, which is when all the workers in a sector can bargain with all the bosses in a sector, which is how you get these freelance writers who don't work for Warner, who still can unionize and demand that Warner give them a certain contract, even though they're not even employees. You know, I think there are a lot of my peers these days who think that the answer to AI training and our bosses wanting to erode our wages with models is a stronger copyright, the thing our bosses say they want. But, you know, we've got a lot of experience with this. We've had stronger copyrights regularly introduced for the last 45 years. They've made the companies we work for the most wealthy and powerful media companies the world has ever seen. And we get less money than we ever have. And that's because when you bargain. with five publishers or four studios or three labels, you don't bargain with, you bargain away. Any new right you're given by Congress, they just take. And, you know, the writer's strike was not a strike about the fact that Michael Eisen, or not Michael Eisener, Michael Iger, Bob Iger, beg your pardon, Bob Iger and David Zatislav and the other studio bosses wanted to use chatbots so they could stop paying writers. The strike wasn't about the, you know, a ban on the studios, uh, material being used to train a chat bot. The studios want the material being to be used to train a chat bot. They just want to be paid when it happens. So really, you know, the studio heads and the Sam Altman's of the world are, are planning to eat us all up. And they're having an argument about who should get the biggest bite.

I just don't want to be eaten. And I think that the writer's strike showed us that a labor right can protect you from this where a copyright has failed and failed and failed and failed for the last 40 years. So to condense it, the difference between centerhood and reverse centerhood, at least when it comes to the workplace. is power. Absolutely. Well, and it's not a coincidence that modern economics denies that power exists, right? You know, modern economics is the search for a way to take all of the interactions that we have in the world and reduce them to equations, which means that they can only deal with quantifiable aspects and the qualitative aspects either need to be jettisoned or quantized with some large loss of fidelity. And, you know, as a result, we get these things like revealed references, which is a bedrock of modern economic doctrine that says that when you see someone do something that's different from what they say they want, you should assume that they are wrong about what they say they want and that their true desires are being reflected in their actions, which is the thing you can only believe if you've undergone the very specific neurological injury that you get when you take an economics degree, which is the injury that makes you incapable of either reasoning about or perceiving power, which is how you can see someone who sold their kidney to make the rent. and say, hey, look at that person with the reveal preference for only having one kidney. I do outside the workplace, I struggle with the sharp delineation between centaur hood and reverse centaur hood. Because when a technology becomes dominant, it shapes the institutional arrangements around it. I think of a car. You're a centaur when you're driving a car. The machine is working for you. You're traveling faster than you could on your own two feet and you can haul stuff. But when the car is a hegemonic technology, you don't get a choice about being stuck in traffic on a freeway going through sprawl because the entire transportation and development system has been reshaped around the dominant technology. Yeah. Well, look, the decision that the internal combustion engine should be primarily used for personal vehicles and not for buses or trams or taxis or. you know what have you or that cars should be the the uh have priority on our roads those are all political choices there's nothing inevitable about them they were made by dominant firms uh and those dominant firms mobilize the capital they got from their dominance to capture their regulators and uh and to you know create ideas like pedestrians which was not a thing we had before the the rise of the car which, you know, obviously you think about it for 10 seconds. It becomes very obvious that why would you have a word for someone walking in the road where everyone walked? That's just a person, not a pedestrian. Pedestrian was invented as a category to blame people for when they got hit by cars because you were a pedestrian in the road, and the road is not for pedestrians, it's for cars. But the roads were there before the cars were there. And so these are, again, political choices. You know, the public transit that dominates most cities is buses. They all have internal combustion engines. The decision that most of our ICE should be in private vehicles and not in those buses, that's choice. There's nothing about the ICE that says that that's the way it should be deployed. I had a few questions about your other area of work, which is in science fiction. Because it seems to me you can't separate the AI hype cycle.

from the way we have been primed by popular culture to think about AI? Like the fact, the fact that people with pitch deck started calling machine learning artificial intelligence means they were tapping into archetypes like SkyNet. and Hal and The Matrix, right? Do you think dystopian sci-fi kind of set us up for enriching some of the worst people in the world, making promises they can't come through on? I mean, I can't deny that it's played a role. I think it's remarkable that it has because it's so obvious to me as a science fiction writer, I've written dozens of novels that... the point of science fiction is not to predict the future, but to comment on the present using futuristic parables, you know, a technique that is just like blindingly obvious in how it works. I mean, you look at our most famous science fiction novels in 1984, which is a novel about 1948, a thing that George Orwell made no secret of, right? It was just very obvious what he was writing about, or even Mary Shelley and Frankenstein, the first science fiction novel, which she was very clear was a novel in support of the Luddites and a novel. in support of the idea that we shouldn't let entrepreneurs decide how technology is going to be unleashed on our society, that the principle of co-determination that had been enshrined in law before the Industrial Revolution and was trampled by the mill owners, that that was the basis on which we should have proceeded and that we should proceed in future. These are just obvious things. And the fact that, you know, William Gibson has to go around and tell people, hey, cyberpunk was a warning and not a suggestion is... you know, baffling. That viral tweet about not creating the torment nexus has an unfortunate amount of truth in it. And, you know, to be fair, I do have some colleagues in my sector and my industry who think that they're fortune tellers. I find them extremely cringe, but they exist and they insist that they are seeing the future. But Honestly, I can't think of anything more dismal than the idea that you can predict the future. I mean, if you could predict the future, then it would mean that it was coming no matter what we did, which would mean that what we did didn't matter, which would mean that there's no reason to get out of bed in the morning. You know, I believe in a shapeable future, not a predictable one. I didn't think it was interesting when when. Virtual reality was kind of the next big thing in tech and tech valuations, that everyone was borrowing their language from snowfall and not from neuromancer. Snow crash, excuse me. Yeah, poor Neil. We actually, I was touring with one of my science fiction novels. I did an event in Seattle where Neil was kind enough to be my interlocutor. And we commiserated and exalted in having both coined words, I, in justification had just been adopted into a few dictionaries. And the Metaverse, obviously, his word has had been used as the base. for a $61 billion stock swindle by Facebook. And it was it was quite a remarkable time to be a coiner of words. Do you think that the dystopian streak in popular sci-fi, particularly Hollywood sci-fi, is what has made it?

possible to inflate the bubble we're experiencing now? I think there's something to that. I mean, I do think that back to leave insulin, his notion of crita hype, you know, when you insist that the AI is so powerful could bring about the end of humanity itself, you are intrinsically saying it's so powerful that any other claim you make about it is probably true too. And so there is an extent to which there is symbiosis between a certain kind of AI critic, what are often called the AI safety people, and the AI boosters who they claim are kind of their sworn enemies, because, you know, I have built God, look upon my works, and he has built God, look upon his works and do something about it, are both contain. the stipulation that that person has built God. And I do think that those narratives are definitely assisted by the, you know, cultural inculcation of the idea that an AI could be a runaway intelligence. Although as we see, wherever we see runaway AIs or runaway machines in science fiction literature, they are always stand-ins for runaway corporations. Right. And runaway greed and runaway, you know, like and runaway industry. They're never like the autonomous, immortal life form that uses human beings as gut flora is not AI. It's a limited liability corporation. And if you're Elon Musk, you're the richest man the world has ever seen. You get it to bed every morning. You try to tell your corporation what to do and it doesn't happen. It can be very easy to assume that. what you've got is you've created an autonomous entity that is like a new species and more powerful than you and that you can no longer direct, right? It really does make the AI story particularly resonant, I think, with those people specifically. You've written about how the urge to pitch AI as a replacement for all things human, separate from... the imperative to just hype a technology to inflate a company's valuation seems to reflect a fundamentally solipsistic urge among the elite people who are getting rich. Yeah, look, hell is other people. Sartre wasn't wrong. The reason hell is other people is that other people are great. And when you can figure out how to make it work with them, you can do so many wonderful things. You can literally become superhuman, which is to say you can do more than one person can do, which is what superhuman means. And yet, the problem is that other people just stubbornly refuse to acknowledge that they should just do what you tell them to do. and they should organize their affairs around your priorities and not theirs and so that makes things very difficult and you have to figure out how to cajole them or coerce them or bargain with them or pay them or compromise with them in order to get stuff done and so i think all of us at one time or another have wished that we weren't so dependent on other people because

when it works with other people, it's so great and it's so hard to make it work. And I think billionaires particularly are in that position as our bosses, right? That they have all the power in the world and yet they can't make the things that they want happen. They have these ego-shattering confrontations with people who know how to do things but won't do what they're told. When I was out on the picket line with the writers in L.A., one of them told me that you give notes to a writer's room the same way that you give prompts to an LLM, you know, make me... E.T. But it's about a dog with a love interest and a car chase. And admittedly, that's just air bud. But mostly when you ask writers to write that for you, they roll their eyes at you and say, hey, like. the grownups are making a movie, go twiddle your spreadsheets, you know, and leave us alone. And I think that, you know, this gives rise to the fear that although you think you're in the driver's seat and you fancy you're in the driver's seat, you know that if you don't show up for work, everything is fine. And if they don't show up for work, everything shuts down. And so... Maybe AI is the way you wire the toy steering wheel that you're actually in the back seat with directly into the car's drive train so that you can, you really can be in the driver's seat. And you see this repeated up and down the stack with AI. You know, Mark Zuckerberg managed to trap you on Facebook through what's called the collective action problem, which is... that you love your friends but they're a pain in the ass and you can't agree on when it's time to leave facebook and as long as you love them more than you hate him he can make your life miserable with ads and you'll stick around the problem is that your friends are people and hell is other people and these other people refuse to optimize their friendship to maximize the number of ads you see on facebook and so he replaced those other people largely with theater kids who performed amateur dramatics and uh those people will spend every hour that God sends trying to figure out how to get you to stay on the platform, provided that he dangles remarkably small sums of money in front of them. But they are also other people, and they want equitable treatment and fairness and so on. And so now he's proposing to replace those people with chatbots, and he wants to build a social media service that doesn't have a socializing on it in the same way that your boss might build a workplace without any workers in it, or that, you know, Donald Trump, when he fires all the civil servants with Doge, is dreaming of a civil... civil service without civil servants, politics without bureaucrats, where you just make an order, and there is no so-called deep state, which is to say people who know how things work, who tell you that it's illegal or impossible or immoral or what have you. Corey, Dr. O, I know your time is tight. Thank you. Thank you. Corey, Dr. O is a prolific blogger, digital rights activist employed by the Electronic Frontier Foundation, and the author of many books. Much science fiction, but also quite a bit of nonfiction. His most recent on that front is the reverse Centaur's Guide to Life After AI. If you would like a copy of this book, it is not too late to contribute to KPFA's Summer Fund Drive. We wrapped up the formal fundraising about three days early, which is amazing.

Thank you so much to everyone who chipped in to make that possible. We did end about $6,000 short of our goal. That's because we have to make a call about 48 hours ahead of time when we're going to end a fund drive. So we take our best guess of what amount we'll be able to raise by what time of day. We never quite precisely hit the mark. At this moment, we have a shortfall that it is more than possible for us to make up. And we'd really like to do it because it validates the station's efforts. to rein in the length of our fund drives to make them less intrusive on the coverage that you tune in for. So if you didn't get a chance to give, and if you would like to encourage the movement towards shorter fund drives, you can also pick up a copy of the book we were describing. It's available through our website when you donate at kpFA.org.

All right, that does it for Upfront this morning. We are weekdays at 7 a.m. We post information about our topics and guests at KPFA.org. If you'd like to send in a question for Monday's Corona Calls segment with Dr. John Swartzberg, send that email to Corona Calls at KPFA.org. A big thanks to our producer, Kimya Akbri, our engineer. Kirsten Thomas, possibly Rod Akeel today. My name is Brian Edwards Teakert. Thanks as well to Max Pringle for the news headlines. Remember if you're incarcerated and postal mail is an easier way for you to reach out. You can always send a letter to Upfront, KPFA, 1929 Martin Luther King Jr. Way, Berkeley, California, 94704. Stay up. Stay tuned. A rude awakening with Sabrina Jacobs is coming your way in just over two minutes at eight.