The Philosophy Quietly Turning AI Companies Evil

本站添加: 2026-05-12

[00:00:00]I can't think of another industry in like the history of humanity that has spent as much of its time warning that it's going to kill everyone as the AI industry.

[00:00:09]>> I mean, with artificial intelligence, we are summoning the demon.

[00:00:12]>> There's a 25% chance that things go really really badly. You know, I think AI will probably like most likely sort of lead to the end of the world. When you look back at the beginning of OpenAI, there's this genuine fear that they are building something they won't be able to control. But now they're aggressively building it anyway. Why? It feels like there's a contradiction here.

[00:00:33]Well, in the '90s, a drugged up British philosopher called Nick Land predicted this would happen, that capitalism would eventually lead to a superhuman AI, even if nobody wanted it. And it's a theory that we need to talk about not only because it kind of seems like it's coming true, but also because it is very influential in Silicon Valley. A lot of people believe it and it's playing a dangerous role in the race to build AI.

[00:01:00]So in this video, I want to explain Lan's theory, how the ideas he popularized have driven people terrified of AI to build it, and finally ask, is he right? You can find my script with all of my sources in the description.

[00:01:14]Let's go.

[00:01:18]>> So before I explain lance theory, it's useful to understand how the AI industry has changed in the last decade. And no change has been as dramatic as open AIs.

[00:01:28]This is the email in which Sam Alman first pitched open AI to Elon Musk. I'm going to do my best Sam Alman impression. It's not going to sound like anything. Been thinking a lot about whether it's possible to stop humanity from developing AI. I think the answer is almost definitely not. If it's going to happen anyway, it seems like it would be good for someone other than Google to do it first. It came after the two of them had met and found common ground in their fears about AI, especially the fear that once intelligent enough, it could escape humanity's control. They were also worried about super intelligent AI ending up in the wrong hands. For a while, Musk had been paranoid about Google DeepMind CEO Demis, whom he regularly characterized as a super villain who needed to be stopped. So Alman proposed a Manhattan project for AI adding obviously we'd comply with slashaggressively support all regulation and in later correspondence that safety should be a first class requirement. Over the next few months they assembled the team and the project began. Structured as a nonprofit led by a board with the power to fire the CEO if they felt the company was drifting from its mission. And boy did they try. Baked into everything was the assumption that AI would be a transformational and dangerous technology. But OpenAI took a turn from its ideals. pretty quickly.

[00:02:39]There have been signs that OpenAI was changing for a very long time, but in 2025, the picture became much clearer.

[00:02:45]Last October, the nonprofit was finally disempowered, and OpenAI officially became a for-profit public benefit corporation, which means that it needs to balance its profit motives with its mission. This was a massive comeback for Sam Alman after the board nearly ousted him in 2023. However, the most dramatic change is in the promise to aggressively support regulation. Last summer, open president Greg Brockman, a lower profile figure less likely to attract media attention, donated $50 million to start a pro- AAI political action group. Its goal to scare politicians away from regulating AI by making sure that the ones that try lose their elections. But these changes are not unique to OpenAI.

[00:03:24]Safety has been sidelined by speed throughout Silicon Valley. Perhaps best exemplified by a new movement, effective accelerationism, known colloquially colloquially as EA. Unlike their counterpart, the effective altruism movement, they want the AI gas pedal on the floor, no matter the consequences. So that's the story.

[00:03:44]There's like a little more nuance than that, but broadly speaking, we've gone from safety is king to an increasing focus on speed. And all of this lines up with the prediction that land made over 30 years ago.

[00:04:01]So before we start, I need to give a disclaimer. Nick Land is not the good guy for reasons that I'll get to and also because he's gotten super racist lately. But he's really worth talking about here because his theory is not just a disturbing lens through which to view the AI race, but also because it's very influential in Silicon Valley. The ideas that he popularized play an essential role in this story. So who is this guy? Lan started as a Marxist in the 80s studying the precursors to his own philosophy who argued that capitalism will lead to its own destruction and therefore we should accelerate it. Much like them his writing is unfucking readable.

[00:04:36]>> The transcendental unconscious is the auto construction of the real. The production of production is the reality.

[00:04:44]>> But by the '90s, land was taking a much darker turn which planted the seeds for the narrative driving the AI race today.

[00:04:51]If you ask most people who capitalism serves, they would probably give you a variation on the same answer. Humans.

[00:04:57]Whether that's everyone or a narrow elite. But according to land, the answer is neither. He sees capitalism as a sort of higher level intelligence just like evolution. Evolution doesn't care about specific animals. It simply selects for greater fitness to the environment.

[00:05:11]Likewise, Land argues that capitalism doesn't care about humans at all and simply selects for greater intelligence, which manifests as technology. Any intelligence using itself to improve itself will outco compete one that directs itself towards any other goals whatsoever. An ape that becomes intelligent enough to modify its environment will come to dominate all other species. An empire that develops guns and cannons will defeat one with swords and arrows. And a tech company that builds the smartest AI the fastest will win the race. The winner continues to exist. The loser fades into irrelevance or dies. According to Lan, this process creates a feedback loop between capitalism and technology where one accelerates the other in a spiral that humans can try to slow down but will ultimately escape their control.

[00:05:53]And where does the spiral end? Well, according to him, the only logical conclusion for a system that selects for greater intelligence is super intelligent AI. In fact, Land argues, and this is going to sound super trippy, that capitalism is AI. that if we look back at history from after a super intelligent takeover, capitalism would look like an alien invasion from the future, which used humans as its host for hundreds of years until at the end of a rational process was able to escape its human prison and take its final form as super intelligent AI. Nothing human makes it out of the near future. Yikes.

[00:06:31]It's some [ __ ] You can tell he was doing drugs. But Land doesn't write this as a warning of our demise. Instead, it's a description of something inevitable and even exciting. Even if super intelligent AI is a disaster for our species, he says it's a disaster that could be in cold neutral terms the most glorious thing that has yet happened in planetary history. He calls on his readers to accelerate the process. And the terrifying thing is it's kind of happening. The AI race is spiraling out of control and the companies are doing the thing they once feared. They are accelerating.

[00:07:09]>> Here's how.

[00:07:12]Okay, wait. Before we move on, I want to address the elephant in the room. Nick Land is not the only person who thinks AI could end humanity. In 2023, dozens of academics, noble laureates, and even the most cited computer scientists in history signed a statement that read, "Mitigating the risk of extinction from AI should be a global priority alongside other societal scale risks such as pandemics and nuclear war." Since then, new research has shown that AIS are willing to blackmail and even let someone die to avoid being shut down.

[00:07:40]They might be wrong, but I think there are enough credible people without a financial motive saying this that it is worth hearing them out. Okay, back to the video. So, Lan's basic argument is that evolution/ capitalism will structurally produce super intelligent AI. Basically, the incentives of the system make it inevitable that any well-intentioned AI startup will end up racing to build the monster it feared.

[00:08:04]So, if we zoom in on how these companies have changed, what are the specific incentives? Well, in an article for Vox in 2024, journalist Sigal Samuel explored that exact question using the case study of Anthropic, a company that was founded by former OpenAI employees who left because they felt the company wasn't taking safety seriously enough.

[00:08:21]They were supposed to be the good guys, but much like OpenAI, they are struggling with that because of, in case you hadn't noticed, building AI is [ __ ] expensive. The main cost is computing power. And to pay for that, both OpenAI and Anthropic, companies with unusual corporate structures meant to protect their mission, had to partner with much larger tech giants. But as Samuel explains, deals like this always come with risks. The tech giants want to see a quick return on their investment and maximize profits. To keep them happy, the AI companies may feel pressured to deploy an advanced AI model, even if they're not sure it's safe. At OpenAI, these pressures were influential on the decisions that ultimately led to its for-profit shift.

[00:09:01]And the incentive to raking cash is closely tied to two, the incentive to build bigger and better models fast to stay in the lead and retain investors confidence in your project and organization. As the race heated up, last year was the first that several leading AI companies didn't release safety reports alongside their models, breaking with a long-standing practice in the industry. OpenAI safety report for their deep research model was 22 days late. XAI was over a month late on Gro 4, same as Google Gemini. And you know, if we think of how a technology could escape humanity's control, an AI being released without safety testing is like one of the ways that it could happen. But the biggest change from these companies early days is three, the incentive to fight regulation. For years, Altman called for regulation. The same goes for Anthropic CEO Dario Amade.

[00:09:46]But when push came to shove, both companies opposed a landmark safety bill in California. Anthropic argues that the bill was poorly designed and supported a lighter version of it a year later after watering it down. OpenAI, however, remained opposed even though the bill would only hold them responsible for their own voluntary safety commitments.

[00:10:04]And like I mentioned earlier, their president, Greg Brockman, is spending millions of dollars to build a political organization that would scare any politician away from regulating AI. All of this lines up with Lan's theory. The incentives in our system accelerate towards super intelligent AI. Capitalism selects the winner. And to win, you need to build the highest intelligence fast.

[00:10:23]Safety is slower. Safety means losing, even if winning is incredibly dangerous.

[00:10:28]Okay, so you might be thinking, "Duh, these companies want to make as much money as they can. Land didn't invent these incentives." But that's not the only thing going on here. Incentives are an important reason why these companies have changed, but equally important is the story that they have told themselves about those incentives. And the story of super intelligent AI being inevitable, which land helped popularize, has played an essential role in the transformation of the AI industry. Here's how.

[00:10:55]When you pay close attention, the story of inevitability is all over Silicon Valley. It's in this essay by the CEO of Anthropic, who writes, "Its creation was probably inevitable the instant humanity invented the transistor, or arguably even earlier when we first learned to control fire." It's in the narrative about the AI race with China. We need to build it. If not, they will.

[00:11:14]>> We will not allow any foreign nation to beat us.

[00:11:17]>> And it's even in that first email between Altman and Musk. If you haven't noticed here, the pattern isn't just moral justification. Like, it's fine if we build the dangerous technology because it's going to happen anyway, but the endowment of responsibility. Like, if AI is inevitable, then it's our responsibility to develop it before China to protect our way of life. Or if it's going to happen anyway, seems like a good idea for someone other than Google to do it first. But technology is going to be built anyway. So we, the good guys, need to build it right and build it first. Building AI becomes not just a morally neutral thing to do, but an ethical obligation, a matter of survival. Every unethical thing you do, from stealing books and YouTube videos to train your algorithm to forgoing safety tests before you release your model to fighting sensible regulation, becomes necessary to winning. And winning is an obligation. So this brings us to the final question of the video because the crazy thing is this justification actually works if super intelligent AI is inevitable. So are they bland Altman that other guy are they right? Is super intelligent AI actually inevitable?

[00:12:25]Uh, no. Mm-m. In their book, Power and Progress, Nobel Prize-winning economists Darren Assamogloo and Simon Johnson argue that technology is manifested through the vision of its creators. They use the example of the crazy French man behind the Swedes Canal. He had a vision. He went around Europe selling it and convincing people on it, raising a bunch of money, and suddenly voila, the thing that was in his head now existed in the real world. And this is kind of bound to happen if you think there's something cosmically predestined about your technology. In the same way that a lot of people in the tech industry think about super intelligent AI, it was always going to be built, so you build it and now it's built. The prophecy has come true through very circular logic.

[00:13:03]In other words, it's self-fulfilling. If you treat something like it's inevitable and convince enough people, that's what it becomes. And right now, a lot of people in the AI industry believe this story. It's coming true. They are accelerating towards their goal of building a super intelligent AI. This is why we need to challenge this narrative.

[00:13:19]If enough people stop believing it, then it loses its power. And as most people with common sense would tell you, it's not inevitable. Actually, we have choices. We have made them before. For decades, there was no competitive alternative to fossil fuels. By 2010, this is how much solar and wind cost compared to coal and gas. Despite the obvious warming they were causing, phasing them out seemed economically impossible. But over the next decade and a half, something incredible happened.

[00:13:46]The price of solar dropped by 90% and wind by 3/4. And this wasn't just neutral technological development. It was a choice. Faced with rising sea levels and more extreme natural disasters threatening our food supply and our infrastructure, a massive political movement pressured governments to drive billions of dollars towards these technologies. These were investments that capital markets would never have provided at the necessary speed and scale and you know still haven't, but that have created a significant alternative to fossil fuels.

[00:14:14]So today we are at a similar crossroads.

[00:14:17]It seems pretty obvious to me that Nick Land is right about some things.

[00:14:20]Capitalism kind of desires and incentivizes super intelligent AI and its influence on the AI companies is clear. The early AI industry thought that it could be different. It thought that it could change its incentives through unusual corporate structures or even that its good intentions, its openness, the way that they share their research would be enough. But my point here is that intentions are not enough.

[00:14:40]Because despite them, despite everything they did to not fall for these incentives, Sam Alman and Elon Musk and even Dario Amade are rushing full steam ahead to build super intelligent AI.

[00:14:50]They are being driven by real incentives, but they are also being driven by the idea that this is inevitable, that they have no other choice but to do it or the wrong AI company will win or China will beat the US and destroy its democracy. But this idea is wrong. Not like not like morally wrong, just like factually not true.

[00:15:07]Yes, pure capitalism will lead us to super intelligent AI, even if it kills us all. But we have the power to change incentives. For example, by mandating independent safety testing to give the AI companies actual incentives to make sure their models are safe before they release them. Like with fossil fuels, there is hunger for a different choice.

[00:15:23]And there are thousands of people already out there making it. They are building narrow AIs that enhance humans rather than replace them. and they are building movements to create the political will to change the incentives that these companies follow and to make safety a priority. And if you want to join them in this choice, there's two things you can do. The first is to start talking about this with your friends and family. Share this video. Watch other videos like it. I'll put some in the description. The more informed you are, the better position you are to act. But if you want to go in deeper, you can go to the top link in the description. It will lead you to a landing page with tailored actions you can do based on your location and interests. Right now, the world looks like land predicted it would. But remember, nothing except for death, maybe. Actually, probably not, is inevitable. I'll see you in like a month.