Zuckerberg's AI manifesto reinforces public distrust of tech

On Monday, Mark Zuckerberg published a 6,500-word manifesto outlining his personal vision for AI, focusing on Meta's development of "personal superintelligence" systems. While versions of these ideas have appeared in the Wall Street Journal and Meta earnings calls in recent months, this essay represents his most detailed public articulation of the topic. Rather than a rebuttal to "doomers" (those who warn of AI catastrophe), the piece is primarily a passionate description of how Zuckerberg believes superintelligence will transform society—though in practice, it repeatedly highlights the very risks he is trying to downplay.

The manifesto rests on broad philosophical claims about how broadly distributed power will lead to positive outcomes. Zuckerberg argues that if everyone has access to superintelligent tools—tutors, lawyers, coaches, learning assistants—then competition between individuals and institutions will naturally balance power and improve society. He proposes a freemium model with dynamic pricing for compute, claiming this will ensure universal affordable access while allocating resources to whatever people "collectively find most valuable." On the surface, these ideas sound appealing: more access, fairer competition, better outcomes for everyone.

However, the article's author identifies a pattern of dangerous naiveté in the specific examples. On education, Zuckerberg envisions students with "personalized tutors with a PhD in every subject and unlimited patience." But consumer chatbots—the actual product underlying his vision—already exist and are widely used in schools. The primary documented use is not learning but cheating: students ask AI to do their homework and write essays. There is no robust watermarking system to detect AI-generated work, leaving teachers unable to verify whether assignments are the student's own effort. This is not a speculative future harm but a real-time problem. On law, Zuckerberg imagines universal access to superintelligent lawyers eliminating courtroom inequality. Yet the article notes that AI-powered legal tools might instead add complexity to an already "bureaucratic hellscape," or encourage a flood of frivolous lawsuits that clog the system like spam.

On pricing, Zuckerberg's proposal for a dynamic auction mechanism—where users pay spot prices for compute—mirrors theoretical free-market principles. But every consumer AI product shields end-users from spot pricing precisely because surge pricing creates a miserable experience, especially for tools people rely on for work. The author assumes Zuckerberg knows this and does not intend to implement literal dynamic auctions, but the ambiguity itself becomes a trust problem: readers cannot tell if he is being disingenuous or dangerously detached from how these tools are actually used.

Underlying all this is a credibility crisis. Facebook and Mark Zuckerberg are deeply unpopular in the United States. A recent survey found 64 percent of Americans believe social media has been harmful to democracy, with similar numbers supporting stricter regulation—and these views cut across partisan lines. This past weekend alone, a court fined Meta $567 million for harm to children. Against this backdrop, Zuckerberg's refusal to acknowledge concrete dangers in AI—harms already visible in real classrooms and courtrooms—reads as either willful blindness or a calculated rhetorical strategy to escape accountability. Either way, it fails to rebuild trust. By contrast, other AI leaders like Sam Altman and Dario Amodei have been more effective at public communication by directly acknowledging risks, explaining their own safeguards, and inviting scrutiny. That approach is not foolproof—safeguards sometimes fail—but without it, the industry risks being unable to survive inevitable failures. Zuckerberg's essay, in refusing that harder work, exemplifies exactly why so many people remain anxious and skeptical about AI.