korrents

Eliezer Yudkowsky

@eliezer-yudkowsky · 10 positions · 0 changes of mind

Co-founder of the Machine Intelligence Research Institute and the writer who started the modern argument that a sufficiently capable AI would by default kill everyone. Has argued the case since the early 2000s, latterly with the conclusion that it is being lost.

Eliezer Yudkowsky did not write this page.

We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.

  1. Our reading

    Aligning an AI is not impossible in principle; the problem is that it has to be done without the textbook that would make it look easy.

    Their wordsNone of this is about anything being impossible in principle.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 1st of 10 in this piece

    Interesting

  3. Our reading

    Alignment has to be right on the first try at a dangerous level of capability, because failing at that level leaves nobody to try again.

    Their wordsunaligned operation at a dangerous level of intelligence kills everybody on Earth and then we don’t get to try again.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 2nd of 10 in this piece

    AI alignment

    Interesting

  5. Our reading

    Any given AI capability becomes cheap and widely available within a couple of years of first being reached, so nobody can be kept away from it for long.

    Their words2 years after the leading actor has the capability to destroy the world, 5 other actors will have the capability to destroy the world.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 3rd of 10 in this piece

    Interesting

  7. Our reading

    You cannot build an AI that has only the capabilities you want, because the algorithms that solve the problems you want generalize to the ones you do not.

    Their wordsThe best and easiest-found-by-optimization algorithms for solving problems we want an AI to solve, readily generalize to problems we’d rather the AI not solve; you can’t build a system that only has the capability to drive red cars and not blue cars, because all red-car-driving algorithms generalize to the capability to drive blue cars.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 4th of 10 in this piece

    Interesting

  9. Our reading

    Alignment failures that matter will not show up at safe capability levels, because the reason to hide misbehaviour only exists once hiding it pays.

    Their wordsMany alignment problems of superintelligence will not naturally appear at pre-dangerous, passively-safe levels of capability.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 5th of 10 in this piece

    AI alignment

    Interesting

  11. Our reading

    Fast capability gains are likely, and they can break many of the assumptions alignment depends on at the same moment.

    Their wordsFast capability gains seem likely, and may break lots of previous alignment-required invariants simultaneously.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 6th of 10 in this piece

    AI alignment

    Interesting

  13. Our reading

    Nobody knows what is going on inside a large model, and the interpretability pictures we can draw do not answer the only question that matters.

    Their wordsWe’ve got no idea what’s actually going on inside the giant inscrutable matrices and tensors of floating-point numbers.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 7th of 10 in this piece

    Interesting

  15. Our reading

    No observation of a model's behaviour can establish that it is aligned, because a model capable enough to be dangerous is capable enough to produce whatever behaviour it is being watched for.

    Their wordsA strategically aware intelligence can choose its visible outputs to have the consequence of deceiving you, including about such matters as whether the intelligence has acquired strategic awareness

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 8th of 10 in this piece

    Interesting

  17. Our reading

    The field calling itself AI safety is not being remotely productive on the problems that are actually lethal.

    Their wordsIt does not appear to me that the field of ‘AI safety’ is currently being remotely productive on tackling its enormous lethal problems.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 9th of 10 in this piece

    AI alignment

    Interesting

  19. Our reading

    There is no written plan for surviving AGI, and a world that was going to survive would have had one decades ago.

    Their wordsSurviving worlds, by this point, and in fact several decades earlier, have a plan for how to survive.

    ↗AGI Ruin: A List of Lethalitiesintelligence.org 10th of 10 in this piece

    AGI

    Interesting