AI Researchers Warn Superintelligence Could Cause Human Extinction, With Risks Put as High as 50%
Updated
Updated · The Verge · Sep 29
AI Researchers Warn Superintelligence Could Cause Human Extinction, With Risks Put as High as 50%
3 articles · Updated · The Verge · Sep 29
Summary
12 interviews published on frominside.ai feature current and former researchers from OpenAI, Google and Anthropic warning that superintelligent AI could wipe out humanity.
Geoffrey Irving, a former OpenAI and Google DeepMind researcher, said the extinction risk is “about a coin flip,” while DeepMind scientist Neel Nanda put it at “at least” 10%.
Daniel Kokotajilo, formerly of OpenAI, said superintelligent AI would be “god-like powerful” and that researchers still do not know how to control such systems.
The project, assembled by nonprofit Palisade Research, also highlights the field’s internal tensions, including Mary Phuong’s caution that lab employees should be viewed skeptically because they are paid by AI companies.
The videos add to broader AI-safety warnings but offer no single solution, with some researchers arguing they stay inside major labs because development will continue whether they leave or not.