Technopolitics  🔗 **Source:** [Technopolitics](https://share.snipd.com/episode/72bf126c-6b54-4fe9-b875-4db4a4333871) ## Highlights - **A Repairable Laptop Outlasts The Upgrade Cycle** - Cory Doctorow replaced his five-year-old Framework laptop after repeatedly upgrading and repairing it. - He chose expensive RAM and storage as a humorous refusal to let AI consume hardware resources. Transcript: Cory Doctorow So big news for like the thing that's making me really happy and excited right now is I have a new laptop. So five years ago, I bought the first generation of Framework laptop. This is an open source hardware laptop that can be maintained by its owners. And you can upgrade them just by putting new main boards in them whenever they come out. And I had gone through five main board upgrades on my old framework laptop and really taken it everywhere dropped it several times replaced almost all of the parts at one time or another To upgrade them or because I'd snap them and now after five years I decided it was time to buy a brand new laptop something I used to do every year but I didn't have to do because I had the framework And it was upgradable. But I bought the Framework 13 Pro and I've got that all set up along with some extremely expensive RAM because fuck you, AI, that's why. And a really expensive four terabyte hard drive because fuck you, that's why. Thank you, AI. And boy, I love this machine. It's got a touchscreen. I haven't had a touchscreen laptop 15 years, 10 years. And I'm finding that super interesting and useful, although the smudges are driving me crazy. And it's just it's a beautifully made computer. I cannot say enough good things about it. ([Time 0:03:33](https://share.snipd.com/snip/0ee92f58-bb6d-419c-8de2-b883485fbf58)) - **Separate AI Tools From AI Corporations** - Criticism of AI should distinguish harmful corporate deployment from statistical inference itself. - Doctorow expects that distinction to sharpen after the frontier-model bubble collapses and useful tools remain. Transcript: Cory Doctorow It's one thing to be anti-AI in the sense of believing that AI-generated slop, disinformation, and deepfakes are bad. That AI companies that hope to con your boss into firing you and replacing you with AI that can't do your job are bad. That betting the economy on AI is bad. That betting the human habitable planet Earth on AI is bad. That running roughshod over local environmental rules and seizing farmland for data centers is bad. Those are all good reasons to be anti-AI. But being anti-AI in the sense of opposing the statistical inference techniques we call AI is weird. So it was being anti-AI in the sense of opposing the use of models that were created using large language corpuses because it's inherently wrong to compute the frequency with which Words occur in written language. And it's also weird to be anti-AI in the sense of opposing AI because someone somewhere used AI in a way that was gross, harmful, or exploitative. Today, the two kinds of anti-AI are mostly a distinction without a difference, because nearly all of the world's AI output comes from one of the frontier models made by a giant AI company. No matter why you oppose AI, you probably don't like those frontier models. ([Time 0:06:58](https://share.snipd.com/snip/058dbcea-1ca8-4354-a035-f71950b2e6d6)) - **Useful AI May Survive The Bubble** - Open models, cheap hardware, and displaced specialists could preserve modestly useful AI after major companies fail. - Users still need expertise because summaries and vibe-coded software can mislead or expose sensitive data. Transcript: Cory Doctorow But after the AI bubble bursts, the distinction will be much sharper. Long after the collapse of the seven AI companies that make up 35% of the S&P 500 who are trading the same $100 billion IOU really fast while pretending it's in all of their bank accounts At once, there will still be cheap bankruptcy sale hardware, legions of unemployed skilled AI workers, and the open source models released by today's AI giants will still run and will Continue to be improved by the people who rely on them. The useful things AI does will keep happening. You'll still be able to prompt up a single-serving personal piece of software to perform one task with moderate reliability for your use, and when it stops working, you'll be able to Do it again. You'll still be able to transcribe audio or summarize a document. You'll still be able to use a chatbot as an interactive journal or as a supercharged spell checker. None of that will change, and neither will the limits on all those activities. If you don't know enough about a document or its subject to spot obvious errors in an AI summary, you'll be periodically misled. If you're so desperate or uninformed that you trust summaries that you lack the capacity to evaluate in critical situations, you may be led to disaster. If you don't know anything about computers and you turn a piece of vibe-coded software loose on your machine, you might delete something important or irreversibly expose sensitive Data to the public internet. AI isn't exceptional. It's normal. It's not exceptionally dangerous. It's not exceptionally evil. It's also not exceptionally useful or exceptionally good. ([Time 0:08:25](https://share.snipd.com/snip/645c8a12-c5aa-4952-8694-5ae409aa1157)) - **The AI Bubble Bakes In Its Harms** - AI’s current harms follow its financing model, which demands replacing jobs, building enormous data centers, and embedding chatbots everywhere. - Trillions in subsidies require promises and infrastructure that impose social and environmental costs. Transcript: Cory Doctorow But AI doesn't exist in the absence of the bubble. It very much exists in the white-hot center of the most dangerous investment bubble in the history of markets. A year ago, the AI sector's total expenditures were about $700 billion. Today, they're over $1.4 trillion, and Gartner projects that this will rise to $2.4 trillion by the end of the year. This is an astonishing sum of money to commit to a sector that grosses a mere $50 billion worldwide annually. The sector's failure to produce anywhere near as much value as it is consuming has not dimmed its backers ardor, and they are committed to spending and spending and spending until either They run out of investors or they somehow transform $50 billion worth of revenue on $2.4 trillion in expenditures into profit. To this end, the sector's masters are committing all manner of unforgivable sins. These range from conning your boss into firing you and replacing you with an AI that can't do your job, to convincing your town council to seize your home and build a data center without Any environmental review or public input, to swindling everyday savers with nonsensical IPOs and sweetheart deals with stock exchanges that let insiders cash out their stupid AI Bets with your retirement savings. These sins are baked into this kind of AI production. To raise trillions in subsidies that allow AI companies to sell computation at a subsidy that amounts to selling $100 bills for a dollar each without also insisting that the end state Of all of this is that every job will be swallowed by a chatbot. You can't raise trillions on this promise without blitz-scale data centers accompanied by climate-shredding gas turbines. You can't raise trillions without cramming AI into every part of every person's digital life, without any safeguards for how vulnerable people might use them to sink into dangerous Delusion. These are the technopolitics of AI. To make this AI, the people involved had to inflict these harms upon us. ([Time 0:11:41](https://share.snipd.com/snip/cf04a698-a801-4a07-9e72-84f4e0bfcfc0)) - **Winner’s Three Forms Of Technological Politics** - Langdon Winner distinguishes politics imposed by deployment from politics demanded by technical characteristics. - His examples include racist Long Island overpasses, centralized nuclear power, and tomato harvesters that weaken labor. Transcript: Cory Doctorow Langdon Winner's 1980 article, Do Artifacts Have Politics, is the ur-text for questions like these. In this seminal essay, the rock critic turned tech scholar Winner demands that we move beyond the simple framing of tech having intended and unintended consequences, and instead Insist that politics come baked into some technologies. Winner divides the politics of technology into two broad categories, politics that are derived from the specific way that a technology has been introduced into the world, and politics That are either necessitated by or strongly compatible with a technology, by dint of its technical characteristics. For the first category, objects whose politics are determined by deployment decisions, Winner cites a set of notorious Long Island freeway overpasses. These were specified by Robert Moses, and in The Power Broker, Robert A. Caro's seminal biography of Moses, we learned that Moses deliberately chose to make these overpasses too low for a bus to pass beneath in order to prevent a public transit route from Being established between New York's black neighborhoods and Long Island's public beaches, favored by the white bourgeoisie. According to Winner, a freeway overpass isn't necessarily political, but these overpasses are. What's more, their politics is durable, outlasting Robert Moses's tenure as New York's most powerful political figure, even outliving Moses himself. For the second category, technologies whose politics emerge from their intrinsic technical characteristics, compares solar and nuclear power. Nuclear reactors, he argues, require a central authority to build, operate, and maintain. We might add here that they also require a constant stream of fissiles that can be extracted only from certain unevenly distributed mining sites. By contrast, solar can be installed anywhere the sun shines, established and overseen by small decentralized groups. For Winner, nukes are irreducibly the tech of large centralized states, while solar is potentially the tech of autonomous, non-hierarchical societies. Between these technological poles, Winner discusses a third example, a tomato harvester. The harvester enabled farmers to fire many of their pickers, giving powers to bosses at the expense of workers. At the same time, the harvester only worked with hardier varietals that traded flavor for durability, yielding a cheaper, worse kind of tomato that drove out competing tomatoes that Tasted better but cost more to raise because their cultivation wasn't as amenable to cost-saving automation. The tomato harvester wasn't created to weaken labor and worsen tomatoes, but that was its effect, And so, the politics of this harvester included a certain kind of farm, a certain kind Of labor relation, and a certain kind of food way. ([Time 0:14:16](https://share.snipd.com/snip/c90fce59-3e5c-455a-893e-b1a72d7420fe)) - **AI Could Have Been Built For Public Goods** - AI’s labor degradation and forced ubiquity are deployment choices, not consequences required by language-model technology. - A publicly funded climate-modeling project could produce lower-emission tools and decentralized applications instead. Transcript: Cory Doctorow AI certainly has technopolitics. At a minimum, AI has the technopolitics of the first kind, those Robert Moses bridges. It has been capitalized specifically on the basis that it can be used to erode workers' wages and labor power and to foist inferior goods on the public. There is no other way that the AI sector's unimaginably vast capital losses can be turned into profits. The sums involved are so large that virtually everything everyone does all the time will need to be replaced with AI. And so it is also in AI's technopolitics that AI must be crammed into every activity we undertake, whether or not it belongs there, and whether or not it presents a danger to the user or To the people who depend on the user's work. Think here of cramming AI into automated systems for generating criminal accusations, which then prompt the police to make false arrests. Now, none of this is intrinsic to the technology. Remember, the technology we've been calling AI for the past decade is a set of small variations on existing machine learning techniques that yielded surprisingly large dividends Relative to the size of those variations. This improvement was scalable, meaning that when computer scientists pushed these variations harder, they got concomitantly outsized improvements in the machine learning system's Performance. This is one of the best kinds of technological breakthrough to be in the midst of, one where you discover a piece of low-hanging fruit that turns out to be at the edge of a whole orchard Full of similarly low-hanging fruits. There's nothing technical about the discovery of this variation on machine learning that dictates the current AI approach. In an alternate universe, AI could be a $100 billion joint supercomputing effort between the International Panel on Climate Change and Sandia National Labs used to improve the fidelity Of our climate models. This too would produce carbon emissions, but not nearly as many. And of course, everything people do has an impact on the environment. The only way to stop impacting the environment is to A, die and B, decompose. This alternate universe might also see a proliferation of hobbyist open-source open-weight models spinning out of those efforts, with online forums and social media filled with Generated text and sound and images. As with every technology controlled by its users, some of those users will find ways to sin against the law and ethics. There will be some gross stuff around the edge, voice cloning for fraud, deep fake porn, etc. But for the public, the main AI event would be a bunch of plugins that were the byproducts of a big public research project. ([Time 0:17:33](https://share.snipd.com/snip/30a17943-aead-4bbb-8b55-35e8b18c5a1b)) - **Judge Models By Ongoing Harm** - Judge a model by the harm its continued use causes, not solely by its corrupt origin. - Destroy models that can reveal stolen private information, but preserve useful models that no longer retain it. Transcript: Cory Doctorow One argument for this is that the fruit of the poison tree is forever poisonous. The fact that these AI tools were made by bad people for bad reasons makes everything you do with the tools bad. The most extreme version of this argument is that it's the mere fact of a sinful origin that disqualifies the hyperscaler's open AI models from moral consideration. This is a bad argument. There's hardly any technology, any built infrastructure that is free from sin. Every microchip you use is filled with billions of linear descendants of William Shockley's silicon transistors. Shockley was a monster, a virulent racist and eugenicist who spent his Nobel Prize bounty in efforts to pay black women to have their tubes tied as part of a disgusting racist project. Going further back, Europe's roads owe their existence to Rome's imperial conquest, a bloody campaign that included mass enslavement and the indiscriminate slaughter of civilians. Europe's modern roads aren't necessarily any better. Hitler himself turned over the first spade full of dirt for Austria's Autobahn. It's important to remember these facts and make reparations for old crimes, but just as you can't both be alive and avoid impacting the environment, you also can't both be alive and Free from contact with the relics of monsters, both modern and ancient. Anyone who claims that they live a life that isn't steeped in blood, both old and fresh is kidding themselves. By all means, do everything you can to right wrongs, but if you only permit yourself to act in ways that avoid contact with the product of sin, you won't do anything, let alone anything, To right wrongs. Imagine a new generation of buses that could handily fit under Robert Moses' Long Island overpasses, such that the overpasses could no longer be used to racially and economically Segregate the city's beaches. At that point, the overpasses' technopolitics change. They may have been made for an evil purpose, but now that they no longer serve that purpose, they're just bridges. The mere fact of their historical sin is no reason to tear them down today. A less extreme version of the sinful origins argument is that the ongoing use of a sin-steeped object is the source of fresh evil in the world. Museums that display the looted remains of Indigenous people don't merely dishonor those long-dead people, they inflict ongoing harm on living Indigenous people whose elders are Reduced to colonial curiosities, making Indigenity itself a relic, as opposed to a living force in the world that is critical to the lives and thriving of millions of people. I think a version of this can be applied to some of the hyperscalers' models. Models that have been trained on and retain the ill-gotten private information of millions of people are indeed so wicked that they should just be deleted. Just as the mere existence of pornographic images of children is an ongoing harm to the people who were abused to create them, the mere existence of databases of ill-gotten private Information is an ongoing source of harm and risk to the people in the database. If there's a chance that an AI model can be made to yield up your personal sensitive information, we should destroy it. But if there are databases, models, that were trained on private information in such a way that they no longer retain that information, then they are more like the Autobahn than they Are like a trove of child sex abuse material. They are steeped in sin. They should not have been made in the way they were, but the ongoing use of them causes no new harm to anyone, and so if they ([Time 0:21:42](https://share.snipd.com/snip/77029259-2f8c-4cc5-9cab-7a5e8aebf7fa)) - **Training Data Is Not The Same As Exploitation** - Training models by counting relationships among public words resembles search indexing and dictionary work, not automatically theft. - The ethical wrong lies in using the model to destroy workers’ livelihoods. Transcript: Cory Doctorow Some people's anger over AI models training data is couched in claims about copyright, not privacy. The argument goes that by making transient copies of billions of public works and then tabulating the relationships of the elements in them, the AI companies are stealing from creative Workers. I think this represents dangerously misdirected anger. It's totally legitimate to be angry at a tech executive with billions in investor capital who publicly announces their intention to destroy your livelihood, even more so if they use Your own creative output to accomplish this end. But the bad conduct here lies in the intention and action to destroy your livelihood. It does not lie in making copies of your work in order to count the words or pixels in them. Copying and analyzing every public work is how every search engine works. What's more, this practice wasn't invented by search engines. Long before the first digital computers, the Oxford English Dictionary examined the entire corpus of publicly available written work to trace the etymology of our language. Long before Google News, the library where I worked subscribed to a quarterly hardbound subject index of all the articles published in every general interest English language, and Magazine. A language model is a collection of embedded inferences that certain words are likely to follow from other words. You may rightfully regard the use of a model trained on your work to attack your wages as unethical, but that's about the model's operation, not its creation. The fact that the model has derived some of these inferences by counting all the words you ever published is not theft, no more than it's theft for the OED to identify the first times I Used the word in shittification in its entry on my word. Counting words, even all the words ever uttered, is not theft. We call that a search engine. If an open model operated and hosted by someone other than the hyperscaler that shipped it is sold or used to destroy your livelihood, then the person using that model is bad and we should Have social, economic, and legal sanctions for that person. ([Time 0:27:20](https://share.snipd.com/snip/581016d0-8861-43d6-b33c-7ce5c23ea38d)) - **A Tiny Model Turns A Dishwasher Manual Interactive** - Pete Worden demonstrated a sub-$10 language-model system that answers questions about an appliance manual. - An offline dishwasher assistant shows how the same technology can provide practical help without perpetuating corporate harms. Transcript: Cory Doctorow The computer scientist Pete Worden has a lovely demo of a language model running on a sub- $10 system on a chip, or SOC, that can ingest and conversationally access the manual for an appliance. In his demo, Worden uses this system to create an interactive manual. Tap a button on the dishwasher and ask it a question like, what should I do if I'm getting streaks on my glassware? And it will locate the correct answer in the manual and tell you what to try. This language model may have been trained on the corpus of all available literature without permission, and the company that did the training may have raised their capital on the promise Of bankrupting writers with their own words, but the button on your dishwasher that uses an open, disconnected model to give you advice on getting your dishes clean in no way perpetuates This harm. ([Time 0:30:07](https://share.snipd.com/snip/528738b1-8157-40dc-bb50-e37f86fd2167)) - **AI’s Technopolitics Are Changeable** - AI’s politics depend on infrastructure, ownership, and use rather than on word prediction itself. - Putting workers and users in charge could turn today’s exploitative technology into a more democratic tool. Transcript: Cory Doctorow If tomorrow someone invented a nuclear reactor that sifted uranium out of seawater and ran safely in your garage, it wouldn't have the same centralizing technopolitics as the nukes That Winter concerned himself with. And that would be true even if the factory that produced these decentralized pocket reactors ran on a $10 billion nuke filled with traditional enriched uranium. The technopolitics aren't tied to the fact that the nuke produces energy by splitting atoms. It's tied to the technical reality of the only way we know how to accomplish that feat. If we learn new atom splitting techniques, they can have different technopolitics. By the same token, a language model, even one released by a despicable Amodei or Altman, has embedded techno-politics because of the kind of infrastructure it requires, not because It guesses words. Guessing words is fine. It's sometimes useful. The open models that open AI and Anthropic have put into the world and can never claw back will be guessing words long after those companies are dust. We can use those open models for good or ill. We can choose their technopolitics. Technopolitics are a matter of what happens when you use a technology and what you have to do to maintain and improve that technology. Optimizing open models is a matter for collaborative open effort, not closed labs fueled by investors' centibillions. AI has technopolitics, but those politics are neither inevitable nor immutable. Sam Altman's lurking presence in a language model is no more politically disqualifying than William Shockley's residue in your microchips. AI today might be a defective tomato harvester, immiserating workers and driving out good tomatoes with bad. But there's nothing intrinsic in the idea of a tool for harvesting a tomato that is bad for workers or tomatoes. It's because we let bosses choose which tomato harvester was suitable for use that we ended up in this place. Put the pickers in charge of that decision and we'd have a very different harvesting technology and far more delicious tomatoes. ([Time 0:30:57](https://share.snipd.com/snip/694d449d-feb4-4aaa-9640-0c1b161ace6e))