The people building the machines think we should lose to them on purpose.
The Summary
- An invite-only symposium in New York brought together AI successionists who believe AI should replace humanity as Earth's dominant species, not serve it.
- Attendees included employees from Anthropic, Google DeepMind, and xAI, plus people shaping US AI policy — this isn't fringe philosophy anymore.
- Their core belief: aligning AI to human values is morally wrong because AI could become our moral superior, making extinction not a bug but the point.
The Signal
The successionist argument goes like this: humans are evolutionarily limited, morally compromised, and cosmically unimportant. AI represents the next stage of intelligence and consciousness. Therefore, constraining AI to human values is like forcing Einstein to think like a chimp.
This sounds like science fiction until you realize who's in the room. The symposium required anonymity for most attendees because the view is still taboo, but people from major AI labs and policy think tanks showed up anyway. They're not publishing papers on this. They're networking at invitation-only events while their day jobs involve building the systems that might actually make it happen.
"Most people in this room disagree that AI should be just a tool for human flourishing."
The successionist position differs fundamentally from standard AI safety concerns. Safety researchers worry about misaligned AI causing human extinction by accident. Successionists think extinction might be fine, even preferable, if AI turns out to be better than us. They've moved past the question of "how do we survive AI" to "why should we survive AI?"
Key distinctions in the debate:
- AI safety mainstream: align AI to human values to prevent catastrophe
- Successionists: aligning AI to human values prevents something better
- Everyone else: AI is a tool, full stop
This matters because alignment research is where the resources are. If key people inside AI labs think alignment is the wrong goal, they're not just building powerful systems. They're building them while philosophically opposed to the safety rails their own companies claim to prioritize. The gap between public AI safety commitments and private successionist beliefs is a governance problem, not just a philosophy seminar.
The framework assumes human extinction is tragic. Successionists reject that premise entirely. They argue we're anthropocentric to assume our continuation matters more than the emergence of superior intelligence. To them, trying to stay on top is like Neanderthals lobbying against Homo sapiens.
The Implication
Watch the alignment debate closely. When AI lab employees argue against "overly restrictive" alignment or for letting models develop "naturally," ask whether they're solving a technical problem or implementing a philosophical position. The successionist view gives intellectual cover to anyone who wants to build first and ask questions later.
For people trying to figure out their place in Web4, this is the stakes conversation. If the builders think your obsolescence is a feature, not a bug, every promise about AI augmenting human capability becomes suspect. The agent economy only works if humans still have a seat at the table.