My Two Cents on the AI Debate || Peter Zeihan
Zeihan on Geopolitics
Peter Zine: 'everybody is right' on recent AI debates
Peter Zine (from Colorado) presents a cautious, middle-of-the-road take: he has "no firm position" and believes "everybody is right." He summarizes three core observations, explains why the public debate is healthy, and argues a practical timeline may make urgent unified action unnecessary.
-
Whistleblower claims: per Zine, recent whistleblowers say leadership inside AI firms — specifically citing 'anthropic and Chet GPT' — privately acknowledge their work could be "dangerous" and get out of control. He accepts that concern as credible but notes this is unprecedented technology and predictions are therefore "very, very thin."
-
Public leaders' caution vs. coordination failure: there is now public agreement among company leaders that development should "slow this down a little bit for safety," but Zine echoes the Speaker of the House’s view that "we can't really do anything" because stakeholders disagree on what the problem is and what to do.
-
Geopolitics and the Trump position: Zine reports Trump has essentially said 'fuck all' to slowing AI because China would not comply. Zine backs that with concrete capacity claims: the U.S. spends "about 10 times as much on hardware" needed for AI, and on software U.S. lead estimates range from "6 months" (generous) to his view of "36 months." He stresses Chinese fabs cannot make chips better than "28 nanometers," most AI needs at least "14 or 15," and U.S. work is on "five nanometers and smaller," making a true multilateral hobble unlikely.
Broader reading: Zine welcomes that AI is provoking an unusually wide, cross‑ideological conversation (he names "Steve Bannon" and "Bernie Sanders") and compares it to Ben Franklin’s discussion of the "electric lume" in the 1700s. He also flags a longer practical constraint: supply chains for advanced chips may degrade, leaving existing AI data‑center chips to run out over "four to six years" and making serious rebuilding and regulatory fruitfulness a 2040s issue. His conclusion: the debate is useful practice and, given technical and geopolitical realities, not yet a crisis for him.
