A nonprofit just published a dozen video interviews with AI researchers debating whether their own work could end humanity.
Palisade Research, a nonprofit that says it studies AI capabilities and motivations, released the interviews on a site called frominside.ai. The twelve interviewees include current and former employees of OpenAI, Google DeepMind, and Anthropic. Geoffrey Irving, who previously worked at OpenAI and DeepMind, said the chance of human extinction from AI is "about a coin flip." Neel Nanda, a current Google DeepMind research scientist, was more conservative, putting the odds at "at least a 10 percent chance" that AI wipes out humanity.
These aren't outside critics lobbing hot takes. They're people who built the systems in question, on the record, attaching actual numbers instead of vague warnings. That matters because AI risk debates usually stay abstract enough to ignore, and only two of the twelve interviewees quoted here gave specific percentages at all.
Palisade built its case on credibility, not marketing copy, but a slickly produced video series with its own dedicated URL is still a pitch for attention as much as it is a warning.