korrents
Dario Amodei

What Dario Amodei thinks about LLMs

@dario-amodei · 43 positions · On ppll ↗

Co-founder and CEO of Anthropic; previously VP of research at OpenAI.

Dario Amodei 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.

Filter LLMs · Newest first
Topic
All 43LLMs 10AI and jobs 7AI alignment 5data centers 4scaling laws 4AGI 3measuring intelligence 3code generation 2government 2
14 more topics
Silicon Valley 2the cloud 2AI and science 1America 1Anthropic 1benchmarks 1chat interfaces 1China 1coding agents 1democracy 1happiness 1JavaScript 1reinforcement learning 1robotics 1
Sort
NewestOldest

Only Dario Amodei on LLMs. See everyone on LLMs

10 dated positions, 2020 to 2026, in their own words. Our reading of what Dario Amodei has said — not written or endorsed by them.

  1. Our readingModels already generalise substantially from tasks that can be verified to tasks that cannot.

    Their wordsWe already see substantial generalization from things that that verify to things that don't verify. We're already seeing that.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 6th of 29 in this recording

  3. Our readingA lab under real commercial pressure cannot fool itself about whether AI makes it productive, because the output is a model launch every few months.

    Their wordsLike we see the end productivity every few months in the form of model launches. Like there's no kidding yourself about this. Like the models make you more productive.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 9th of 29 in this recording

  5. Our readingLonger context is an engineering and inference problem, not a research problem — nothing prevents it from working.

    Their wordsThere's There's nothing preventing longer context from working. You just have to train at longer context and then learn to to serve them at inference. And both of those are engineering problems that we are working on and that I would assume others are working on as well.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 12th of 29 in this recording

  7. Our readingIf a lab is off by a single year in its revenue forecast, no hedge on Earth stops the compute it already bought from bankrupting it.

    Their wordsAnd if my if my revenue is not a trillion dollars, if it's even 800 billion, there's no force on Earth. There's there's no hedge on Earth that could stop me from going bankrupt if I if I buy that much compute.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 13th of 29 in this recording

  9. Our readingEvery frontier model is profitable on its own; the company loses money only because it is already training one ten times more expensive.

    Their wordsbut at the same time, we're spending $10 billion to train the next model because there's an exponential scale up. And so the company loses money. Each model makes money, but the company loses money.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 16th of 29 in this recording

  11. Our readingModels are far more differentiated from one another than clouds are, so AI will not commoditise the way cloud did.

    Their wordsLike models are good at different types of coding. Models have different styles. Like I think I think these things are actually, you know, quite different from each other. And so expect more differentiation than you see in in um cloud.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 18th of 29 in this recording

  13. Our readingNot every token is worth the same amount, so AI will be sold by results and by the hour as well as by the token.

    Their wordsso so I think we're definitely going to see business models that that recognize that, you know, at some point we're going to see, you know, pay for results or you you know, in some in some form or we may see forms of compensation that are like labor. Um, uh you know, that that kind of work by the hour.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 23rd of 29 in this recording

  15. Our readingTraining a model on principles rather than a list of rules is not an ideology but an empirical finding: its behaviour is more consistent and covers edge cases.

    Their wordsit's it's kind of purely a practical and empirical thing that we've observed that by teaching the model principles, getting it to learn from principles, its behavior is more consistent, it's easier to cover edge cases, and the model is more likely to do what people want it to do.

    ↗Dario Amodei — “We are near the end of the exponential”youtube.com 27th of 29 in this recording

  16. 6 years earlier

  17. Our readingHumans can generally perform new language tasks from few examples or simple instructions, while current NLP systems largely cannot.

    Their wordsBy contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do.

    ↗Language models are few-shot learners (with 30 co-authors)arxiv.org All 3 from this paper

  19. Our readingScaling up language models greatly improves task-agnostic few-shot performance, sometimes matching prior fine-tuning approaches.

    Their wordsHere we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches.

    ↗Language models are few-shot learners (with 30 co-authors)arxiv.org All 3 from this paper