korrents
Rohit Krishnan

Rohit Krishnan

@rohit-krishnan · 22 positions · 0 changes of mind

Writes Strange Loop Canon, on the intersection of technology, economics and business.

Everything they publish, on ppll ↗

Rohit Krishnan 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. The models are pretty gullible and can be induced to do bad things, including and especially by other models

    How to control an agent swarmstrangeloopcanon.com 1st of 3 in this piece

  2. AI agents

    Whistleblower actions don't do anything unless the environment responds to it

    How to control an agent swarmstrangeloopcanon.com 2nd of 3 in this piece

  3. AI agents

    I have said for many years now that the thing I'm worried about with the models is a Black Monday type scenario, where many algorithms work with each other and get us into weird basins of actions.

    How to control an agent swarmstrangeloopcanon.com 3rd of 3 in this piece

  4. 5 days earlier
  5. Cheapness alone doesn't create commodification, what you need is fungibility.

    Who Wins As Intelligence Commodifies?strangeloopcanon.com 1st of 3 in this piece

  6. what this tells us is that the vast majority of the labs' revenue comes from non-frontier models, for which cheaper, comparable alternatives are already available elsewhere.

    Who Wins As Intelligence Commodifies?strangeloopcanon.com 2nd of 3 in this piece

  7. The frontier keeps being extraordinary valuable and that value keeps having a half-life in months or a year.

    Who Wins As Intelligence Commodifies?strangeloopcanon.com 3rd of 3 in this piece

  8. 5 days earlier
  9. AI alignment

    If you do think it's the overall system that matters, then the alignment that's needed is far less like training a virtuous child and more like managing a semi-virtuous corporation!

    AIs are companies, my friendstrangeloopcanon.com 1st of 3 in this piece

  10. Every step we take that makes the models less governable and controllable also makes them less useful, so we will not be able to continue using them, which reduces the value, and which means delaying the release is the only option

    AIs are companies, my friendstrangeloopcanon.com 2nd of 3 in this piece

  11. AI alignmentdesign

    Multi-agent alignment is fundamentally a liberalism project.

    AIs are companies, my friendstrangeloopcanon.com 3rd of 3 in this piece

  12. 5 days earlier
  13. Almost every piece of advice that you hear is basically entirely filled with survivor bias.

    Nobody Knows Anythingstrangeloopcanon.com 1st of 2 in this piece

  14. Ignorance is not a temporary defect that we will eventually eliminate. It's the permanent burden of increasing specialisation and living in a complex world.

    Nobody Knows Anythingstrangeloopcanon.com 2nd of 2 in this piece

  15. 2 months earlier
  16. With people, the problem with committees is that they "smooth out" all idiosyncrasies. They take out any "spiky" points of view, and make things much more normie.

    LLM councils show groupthinkstrangeloopcanon.com 1st of 3 in this piece

  17. peer review

    the council does not simply keep the best bits from everyone. It keeps a minority of the good ideas, while peer review seems to give consensus ideas an extra push.

    LLM councils show groupthinkstrangeloopcanon.com 2nd of 3 in this piece

  18. LLMsAI alignment

    the best way with LLMs usually is to be explicit, since otherwise even if they're aligned they cause emergent problems.

    LLM councils show groupthinkstrangeloopcanon.com 3rd of 3 in this piece

  19. 3 weeks earlier
  20. AI systems however don't need lunch meetings, careers, or envelopes of cash to have an analogous failure.

    Who Audits the Auditors?strangeloopcanon.com 1st of 3 in this piece

  21. audit independence is not a property of the model. It is a property of the system around the model.

    Who Audits the Auditors?strangeloopcanon.com 2nd of 3 in this piece

  22. institutions don't fail because people are wicked. Evidence, incentives, discretion, and review rights might be flowing in the wrong direction.

    Who Audits the Auditors?strangeloopcanon.com 3rd of 3 in this piece

  23. 1 day earlier
  24. benchmarks

    Creating benchmarks is now a job relegated to the smartest and best of us.

    Introducing BenchBenchstrangeloopcanon.com 1st of 2 in this piece

  25. Considering every lab wants to "eat the world" the focus on how to work in real-world messy situations seems apt as their primary home.

    Introducing BenchBenchstrangeloopcanon.com 2nd of 2 in this piece

  26. 4 days earlier
  27. AI agents

    A pile of agents is not a company for the same reason a pile of smart people is not a company.

    Homo Agenticusstrangeloopcanon.com 1st of 3 in this piece

  28. AI agents

    Money might still matter for AI agents because it compresses many messy negotiations into one shared signal.

    Homo Agenticusstrangeloopcanon.com 2nd of 3 in this piece

  29. AI agents

    Central planning is not a solution simply because agents are software, and do not fix the problem of local knowledge which lives near the work.

    Homo Agenticusstrangeloopcanon.com 3rd of 3 in this piece