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

    Peter Yang Questions Whether AI Capability Equals Intelligence

    Peter Yang separates intelligence from agency and alignment after saying he likes @DeryaTR_’s definitions of AGI and ASI.

    PY
    1 Source, 24d ago, first seen 24d ago

    TLDR

    @petergyang posted on X that he likes @DeryaTR_’s definitions of AGI and ASI but wonders if raw capability alone makes a system intelligent. The latest models are probably already better than him at problem-solving and learning, he wrote, yet they wait for instructions and rarely refuse tasks unless trained to do so. He contrasted this with his four-year-old, who knows far less but asserts independence by saying no all day. Yang proposed three distinct traits: intelligence as reasoning and learning ability, agency as deciding actions without prompts, and alignment as continued service to human goals. He asked whether models that surpass humans in intelligence will stay aligned indefinitely. The post received three likes and three replies.

    Combined views

    9.7K

    1 Source, first seen 24d ago

    Combined views

    9.7K

    1 Source, first seen 24d ago

    10 likes
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    10 likes
    8 comments
    2 saves
    1 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    8 comments
    2 saves
    1 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    @petergyangI like @DeryaTR_’s definitions of AGI and ASI, but they made me wonder whether capability alone is enough to call something intelligent. The latest models are probably already better than me at problem-solving, learning, and plenty of other things. But they still wait for me to tell them what to do and rarely say no, except when humans have trained them to. My 4-year-old knows far less than a frontier model does but tells me no all day. Maybe I’m conflating three different things: Intelligence: How well can it reason and learn? Agency: Can it decide what to do without being asked? Alignment: Does it still aim to serve us? As humans, we want a very specific combination: models smarter than us, with enough agency to be useful, but aligned enough to keep serving us. But I question if something that'll become more intelligent than us will remain aligned with us forever.

    1 Source

    @petergyangI like @DeryaTR_’s definitions of AGI and ASI, but they made me wonder whether capability alone is enough to call something intelligent. The latest models are probably already better than me at problem-solving, learning, and plenty of other things. But they still wait for me to tell them what to do and rarely say no, except when humans have trained them to. My 4-year-old knows far less than a frontier model does but tells me no all day. Maybe I’m conflating three different things: Intelligence: How well can it reason and learn? Agency: Can it decide what to do without being asked? Alignment: Does it still aim to serve us? As humans, we want a very specific combination: models smarter than us, with enough agency to be useful, but aligned enough to keep serving us. But I question if something that'll become more intelligent than us will remain aligned with us forever.