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

    Amjad Masad Links AI Risks to Thompson Trust Reflections

    Replit founder shares thoughts on potential poisoned model training.

    AM
    1 Source, 27d ago, first seen 27d ago

    TLDR

    Amjad Masad, founder and CEO of Replit, posted that Ken Thompson’s “Reflections on Trusting Trust” feels super relevant to AI. He described how Thompson bootstrapped a poisoned compiler that left no traces of the poison in the source code because it compiled itself. Masad suggested one can imagine the same thing happening with models where one poisoned generation helps train the next while removing any traces.

    Combined views

    15.1K

    1 Source, first seen 27d ago

    Combined views

    15.1K

    1 Source, first seen 27d ago

    149 likes
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    149 likes
    37 comments
    31 saves
    5 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    37 comments
    31 saves
    5 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    @amasadKen Thompson’s “Reflections on Trusting Trust” feels super relevant to AI. He bootstrapped a “poisoned” compiler that left no traces of the poison in the source code because it compiled itself. You can imagine the same thing happening with models where one poisoned generation helps train the next while removing any traces of the poison.

    1 Source

    @amasadKen Thompson’s “Reflections on Trusting Trust” feels super relevant to AI. He bootstrapped a “poisoned” compiler that left no traces of the poison in the source code because it compiled itself. You can imagine the same thing happening with models where one poisoned generation helps train the next while removing any traces of the poison.