Thread

SNL Weekend Update: Anthropic CEO Dario Amodei on A.I.'S Threat to Humanity [video]

197 points · 105 comments · CharlesW

jihadjihad · 1790536803 · focus · HN ↗
I know it’s a little OT, but I keep seeing the same misquotation all over the place. Right at the beginning Michael Che refers to the claim that “there’s a 10% chance that AI could wipe out humanity within 10 years.”
But that isn’t what the often-misquoted current and former Anthropic employees said. They said it was a 10% chance of it happening by the end of the current decade [0], i.e. by 2030, not 2036.
0: <a href="https://www.theguardian.com/technology/2026/sep/09/anthropic-researchers-ai-human-extinction" rel="nofollow">https://www.theguardian.com/technology/2026/sep/09/anthropic...
1. pj_mukh · 1790536884 · focus · HN ↗
  But what does it matter? It’s not like that number is based on any science. It’s literally p (doom | vibes).
  1. someothherguyy · 1790537135 · focus · HN ↗
    what does the scientific literature say then?
    pj_mukh · 1790537377 · focus · HN ↗
    Unfortunately, this question was never asked by any news organization or of Coxon as he ran to the press. People who would usually ask to "see the work" never did because it comfortably confirmed their biases.
    Specifically, I have no idea what the research says, or even if it exists, I wish we had a system to figure that out instead of exciting Twitter "whistleblower" threads.
    ragequittah · 1790572787 · focus · HN ↗
    I feel like if the doomers are correct and AGI happens research can't say anything about it. We won't be able to predict the outcome of something more intelligent than us or that can disrupt society so thoroughly. It's already causing major problems.
2. nmca · 1790537183 · focus · HN ↗
  You are wrong. The confusion arises because Coxon said “within the decade” with no associated probability and Hubinger said “within the next decade” at 10%.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.