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

    The competing futures behind AI predictions that seem to have aged well

    A user cites 2019 predictions about AI scaling, jobs and capability growth that they say could all fit today's landscape.

    rohitRO
    1 Source, 2h ago, first seen 2h ago

    TLDR

    A user argues that several 2019 predictions could have anticipated today's AI, despite pointing to different futures. They suggest judging a prediction not just by its accuracy score, but by how many possible outcomes would have made it look right—and whether the future it now proposes fits what it predicted.

    Combined views

    734

    1 Source, first seen 2h ago

    Combined views

    734

    1 Source, first seen 2h ago

    7 likes
    7 likes
    1 saves
    1 saves
    Featured Source

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    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

    rohit@krishnanrohitPredictions from 2019 that would've stood the test of time quite well. 1. Scale works for general intelligence, one architecture eats ev'thing 2. Scale teaches any patterns inherent in the data, including patterns of learning 3. AI training will grow but fall prey to Moravec-ish paradoxes 4. AI will zeno's paradox its way through most jobs, displacing larger chunks of it, but not all jobs due to its architecutre 5. We will have slow takeoff in AI capability as linear increase will need exponential investment 6. Anything we can alphazero-fy will see superhuman output, but won't generalise 7. Scaling neuron compute a la Kurzwel will help us get to superintelligent digital minds Note that all of these would have predicted the world of today, but all of these would have different implications for the world of tomorrow, many even competitive. When historical predictive records are being used to support future predictive accuracy it's worth judging not just the brier scores but how many different worlds the predictions would have come true about and whether the future being proposed is congruent with all of them.2h

    1 Source

    rohit@krishnanrohitPredictions from 2019 that would've stood the test of time quite well. 1. Scale works for general intelligence, one architecture eats ev'thing 2. Scale teaches any patterns inherent in the data, including patterns of learning 3. AI training will grow but fall prey to Moravec-ish paradoxes 4. AI will zeno's paradox its way through most jobs, displacing larger chunks of it, but not all jobs due to its architecutre 5. We will have slow takeoff in AI capability as linear increase will need exponential investment 6. Anything we can alphazero-fy will see superhuman output, but won't generalise 7. Scaling neuron compute a la Kurzwel will help us get to superintelligent digital minds Note that all of these would have predicted the world of today, but all of these would have different implications for the world of tomorrow, many even competitive. When historical predictive records are being used to support future predictive accuracy it's worth judging not just the brier scores but how many different worlds the predictions would have come true about and whether the future being proposed is congruent with all of them.2h