Uploaded August 2025 | Updated September 2026, 2 weeks ago
Robert Wright and Nathan Lambert discuss the danger of centralization of power in AI:
"I see the biggest tech companies being so powerful now, and I'm very optimistic about AI's capability,
I think there's a lot to untap, and if it is to make Nvidia, OpenAI, Google, Apple, 20 times bigger, I just don't know what that does to our society, in terms of how much power they have. Yeah, and if you imagine that it becomes like a winner-take-all model, you might not even have three or four companies dominating,
you might just have one. Yeah, the more concentrated AI becomes, the more worries about speech can be there. I think xAI is the leading example where it seems like one sole person has the finger on the voice of the AI, it does this weird Elon search thing if you ask about sensitive topics.
If Grok was a hundred times smarter and everybody would feel a need to use Grok, then it would be a much bigger issue. There clearly are risks about centralization of power, because to talk about freedom of speech almost underdramatizes it. I mean, you're talking about machines that already have pretty impressive persuasive capabilities, and could, in principle, do persuasion on a scale like we've never seen."
Robert Wright and Nathan Lambert discuss the danger of centralization of power in AI:
"I see the biggest tech companies being so powerful now, and I'm very optimistic about AI's capability,
I think there's a lot to untap, and if it is to make Nvidia, OpenAI, Google, Apple, 20 times bigger, I just don't know what that does to our society, in terms of how much power they have. Yeah, and if you imagine that it becomes like a winner-take-all model, you might not even have three or four companies dominating,
you might just have one. Yeah, the more concentrated AI becomes, the more worries about speech can be there. I think xAI is the leading example where it seems like one sole person has the finger on the voice of the AI, it does this weird Elon search thing if you ask about sensitive topics.
If Grok was a hundred times smarter and everybody would feel a need to use Grok, then it would be a much bigger issue. There clearly are risks about centralization of power, because to talk about freedom of speech almost underdramatizes it. I mean, you're talking about machines that already have pretty impressive persuasive capabilities, and could, in principle, do persuasion on a scale like we've never seen."
