6 pointsby hmokiguess18 days ago7 comments
  • tim-tday17 days ago
    LLMS still can’t troubleshoot. I have been thinking about how to solve it and I think an entirely new architecture is needed. Quants got replaced by quant bots the same way human calculators got replaced by electronic adding machines.

    I haven’t met an AI who can think clearly in a systems setting. I’m guessing it might be a decade or two. As an example, ask an LLM to do some 10th grade math. Inspect the thinking process. It can regurgitate the process and the rules but cannot perform them.

    Same with troubleshooting.

    • xeckr16 days ago
      >As an example, ask an LLM to do some 10th grade math. Inspect the thinking process. It can regurgitate the process and the rules but cannot perform them.

      It seems to me that the solution is just RL to get the language model to delegate the actual calculation to the appropriate tool.

    • hmokiguess15 days ago
      I'm not sure your explanation is correct. The quant bots are also decision systems who attempt to predict in non-deterministic environments.

      The reason they got replaced isn't because the problem became deterministic (like a calculator). It's because the error rate and the cost got to an acceptable place when compared with the cost of a human quant.

  • necovek18 days ago
    I think it is never: while we should be able to achieve more with less, smart people show up who can sell more stuff and everyone keeps their job?
  • xeckr18 days ago
    I'll go with <3
  • sleepyguy18 days ago
    No later than 2030.
  • 18 days ago
    undefined
  • randomname432518 days ago
    Companies cut ~600K tech jobs since covid purely on AI hype. Now there is actually (some) utility. The bubble will burst at some point (they always do). I think companies cut to the bone at that point. No one knows when the top is but when it turns I don't think many of the cushy knowledge worker (white collar) jobs come back. It was a good run. 1975 - 202X?
    • clipsy18 days ago
      This seems like a baffling comment. If AI is good enough to replace the vast majority of knowledge workers, why would the AI bubble burst? (Or, more accurately, why would you describe it as a bubble?)
  • Eaglo18 days ago
    [dead]