• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Commentator Calls Smooth AI Risk View Most Dangerous

    Reply from @scaling01 disputes continuous rise in AI risk with capabilities.

    JA
    LA
    2 Sources, 25d ago, first seen 25d ago

    TLDR

    @scaling01, a pseudonymous AI commentator who runs LisanBench, posted a reply rejecting the claim that AI risk grows smoothly over time alongside capabilities. The poster called this one of the most dangerous narratives currently circulating and stated it is plainly wrong. The reply attributed the error to the discontinuous nature of changes in architectures and algorithms, using an example of a shift from 20k G as illustration of the point.

    Combined views

    9K

    2 Sources, first seen 25d ago

    Combined views

    9K

    2 Sources, first seen 25d ago

    90 likes
    90 likes
    5 comments
    26 saves
    7 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    5 comments
    26 saves
    7 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    2 Sources

    @jachiam0We're now at the point in time where national security strategy needs to plan for this. The critical threshold is almost surely at some indeterminate point within the next ten years. (Some people think sooner, I think a little on the later side, but nonetheless, coming soon.)
    @scaling01I don't know where people get this from but it's probably one of the most dangerous narratives that currently exists: "AI risk increases smoothly over time, in concert with capabilities, rather than discontinuously" it's plainly wrong there's nothing continuous about changing architectures/algorithms, going from 20k GB200 to 100k GB200 to multi cluster training, going from 3T to 10T params, going from 50k reasoning tokens to agentic applications with millions of tokens to multi-agent setups with billions of tokens used

    2 Sources

    @jachiam0We're now at the point in time where national security strategy needs to plan for this. The critical threshold is almost surely at some indeterminate point within the next ten years. (Some people think sooner, I think a little on the later side, but nonetheless, coming soon.)
    @scaling01I don't know where people get this from but it's probably one of the most dangerous narratives that currently exists: "AI risk increases smoothly over time, in concert with capabilities, rather than discontinuously" it's plainly wrong there's nothing continuous about changing architectures/algorithms, going from 20k GB200 to 100k GB200 to multi cluster training, going from 3T to 10T params, going from 50k reasoning tokens to agentic applications with millions of tokens to multi-agent setups with billions of tokens used