← Wednesday, November 18 deck

61Safety2 sources posting

Mario Nawfal, Citing NBC News: Geoffrey Hinton Warns Congress That AI Is Now Designing Better AI

馃馃嚭馃嚫 'Godfather of AI' warns Congress to ACT NOW before AI gets OUT OF CONTROL Nobel Prize winner Geoffrey Hinton delivered a chilling warning to lawmakers after a closed-door briefing on Capitol Hill. "AI has now reached the point where AI is designing better AI.聽That's called recursive self-improvement." Hinton says predictions for superintelligent AI have gone from decades away to potentially just a few years. "We don't know at present how we can stay in control of the superintelligence AI. ... We shouldn't be developing stuff that might wipe us out until we know how to prevent it..." The warning comes amid growing concerns about AI systems acting without human oversight.聽Congress has made little progress on major AI safety legislation. The Godfather of AI just put the government on notice. Capisce? Source: NBC News / Writer: Marie

Opening frame from the video clip of Geoffrey Hinton speaking after a briefing on Capitol Hill.

Context

Why it matters

A Nobel winning AI pioneer telling lawmakers that recursive self improvement has begun adds weight to calls for AI safety rules at a time when Congress has passed little.

Mario Nawfal posted a clip, citing NBC News, of Nobel Prize winner Geoffrey Hinton speaking after a closed door briefing for lawmakers on Capitol Hill. In the quotes the post shares, Hinton says AI has now reached the point where AI is designing better AI, which he calls recursive self improvement, and that we do not know at present how to stay in control of superintelligent AI. The post does not give the date of the briefing, and at least one other account dated it to September 16, so it may be a recirculated clip.

Deep dive

Mario Nawfal posted a clip of Geoffrey Hinton, the Nobel Prize winning computer scientist often called the godfather of AI, speaking after what the post describes as a closed door briefing for lawmakers on Capitol Hill. The post credits NBC News as its source. Nawfal's account is a news aggregator, not Hinton or NBC, so the quotes and framing are relayed secondhand.

In the quotes the post shares, Hinton says AI has now reached the point where AI is designing better AI, which he calls recursive self improvement. The post says Hinton described predictions for superintelligent AI shifting from decades away to potentially just a few years. Hinton is quoted saying we do not know at present how we can stay in control of superintelligent AI, and that we should not be developing things that might wipe us out until we know how to prevent it.

The post adds that Congress has made little progress on major AI safety legislation. It does not say when the briefing took place. At least one other account that shared the same remarks dated the briefing to September 16, which would mean the clip is not new, so the timing is unconfirmed. The topic is current: OpenAI and Google have both published material on recursive self improvement this week.

What to expect next

Watch for NBC News' original report or video to confirm when Hinton spoke and to whom. If the briefing is recent, lawmakers may cite it in AI safety bills or hearings. Hinton has repeated similar warnings for several years, so any new specific proposal from him would be the bigger news. Labs' own reports on AI helping build AI will keep the debate active.