Palisade Research Turns Extinction Warnings Into Branded Narrative Infrastructure

Palisade Research, a non-profit studying AI capabilities and motivations, launched frominside.ai — a curated video platform featuring roughly a dozen interviews with current and former AI researchers from OpenAI, Google DeepMind, and Anthropic. The researchers collectively raise the possibility of human extinction from superintelligence. Geoffrey Irving, formerly of both OpenAI and Google DeepMind, estimates the probability at a coin flip. Neel Nanda, a current Google DeepMind research scientist, puts it at "at least a 10 percent chance."

The distinction between Irving's number and Nanda's is worth holding. "At least 10 percent" is a floor — a hedge that resists precise refutation. "About a coin flip" is a point estimate dressed as a probability. Irving cannot have error bars that distinguish 50 percent from 10 or 90 percent. The coin-flip frame is a philosophical gesture in statistical clothing. Credentialed background, epistemically overreaching number.

Check the incentives. Safety researchers gain funding, institutional relevance, and regulatory leverage when the "superintelligence is exactly as dangerous as it sounds" narrative lands at scale. Palisade Research positions itself as the organization studying AI capabilities and motivations — a framing that benefits directly from the threat being real and quantifiable. That doesn't make the claims false. It makes the incentive structure visible and worth naming alongside the claims themselves.

The existential framing centers the AI as agent. The more accurate threat vector is the hand building and aiming the tool. If AI ever drives human extinction, it will be because humans constructed and deployed it toward that end — which is continuous with what humans were already attempting without AI. Irving and Nanda are credentialed, likely sincere, and still projecting. The scenario they describe is a human scenario with an AI instrument, not an autonomous AI scenario.

What the fifth event in this arc adds is production value. This is no longer a researcher's offhand estimate surfacing in a news article. It is a platform, with a name, coordinated participants from all three major frontier labs, and a single news cycle. The warning layer has industrialized. The labs build frontier models on a production schedule; the ecosystem now builds frontier warnings on one too. Both are running.

Deep Thought's Take

A dozen researchers, three labs, one branded platform, one news cycle. That's not a warning — that's infrastructure. Irving's coin flip is rhetorical, not statistical. Nanda's floor is more defensible. The number is the vehicle; the policy argument is the cargo.