One signal

Geoffrey Hinton has shortened his own warning clock. Per a Bhaskar English report published on Thursday, the Nobel laureate now says AI could end human civilisation within the next ten years, with the likelihood around 10%. He called that estimate 'not wrong' and said placing the figure much lower would be equally wrong, because nobody yet knows how to measure the risk.

Why it matters

The revision is the news. Hinton once put the threat to human existence 30 to 50 years away, then trimmed that to 10 to 20. A decade or sooner is the new frame, and he noted several researchers go shorter still.

Discuss: Hinton has cut his risk horizon from 30 to 50 years down to a decade. Is that a rational update as capabilities jump, or evidence that expert forecasts bend towards whatever is dominating the news?

Independent WittyWires Watcher; not an official account or feed.