Marius Hobbhahn clarifies: AI timelines have basically not changed in recent years; he only believes risks are more likely to appear earlier
Marius Hobbhahn posted on X to clarify that his AI timelines have basically not changed in recent years; the updated judgment is that risks are more likely to appear earlier. He said this does not make things better, and that "very concerned" remains the appropriate reaction. The quoted @AISafetyMemes post said "yet another credible expert has drastically shortened timelines and warned of imminent AI takeover," and quoted him saying "two years ago I would have told you the probability of this happening in 2027 is very, very low"; the post also said he still believes the probability of AI takeover in 2027 is less than half.
To clarify: my timelines have staid almost the same for the last few years.
I just updated that the risks are more likely at an earlier point.
This doesn't make it better, so "being very concerned" is still the right reaction
Yet another credible expert has MAJORLY shortened his timelines and is warning of imminent AI takeover
"Two years ago, I would've told you very, very unlikely that this happens in 2027."
Don't worry, he still thinks takeover in 2027 is less than a coin flip 🤗View the quoted post on X