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

    Vedant Misra Posts Bitter Lessons on AI Research

    DeepMind researcher Vedant Misra tweets a list of ways flagship AI projects render other work obsolete.

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

    TLDR

    Vedant Misra, an AI researcher at DeepMind on the Gemini core team, posted a tweet listing ways large-scale AI work overshadows smaller efforts. The post states that flagship jobs become ablations, startups turn into weekend evaluations, products serve as demo notebooks, and research agendas end up as deleted appendices. The message frames these outcomes as the most bitter lesson in the field. The tweet comes from his account and stands as his stated view on current AI dynamics.

    Combined views

    24.4K

    1 Source, first seen 27d ago

    Combined views

    24.4K

    1 Source, first seen 27d ago

    285 likes
    285 likes
    9 comments
    119 saves
    10 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    9 comments
    119 saves
    10 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

    1 Source

    @vedantmisraThe most bitter lesson: your flagship job is their ablation. Your startup is their weekend eval. Your product is their demo notebook. Your ICML oral was their negative result. Your SOTA is their unreported run. Your Series A covers one hyperparameter sweep. Your research agenda is their deleted appendix. Your flagship model underperforms their unannounced prototype. Scaling computation is all that matters.

    1 Source

    @vedantmisraThe most bitter lesson: your flagship job is their ablation. Your startup is their weekend eval. Your product is their demo notebook. Your ICML oral was their negative result. Your SOTA is their unreported run. Your Series A covers one hyperparameter sweep. Your research agenda is their deleted appendix. Your flagship model underperforms their unannounced prototype. Scaling computation is all that matters.