4 ms·

LLMs Are Still Just Predicting the Next Word. So Why Are They So Good Now?

4 points by tudorhn 6mo ago

jqpabc123 6mo ago
Are they so good? They can excel at information retrieval. They often fail at logic and reason. Predicting the next word is basically guesswork. Not the sort of thing any responsible person should rely on to make important decisions. Doing so leads to liability issues in many business and professional environments. The legal system will have significant influence here. https://pub.towardsai.net/the-air-gapped-chronicles-the-court-asked-for-the-llms-reasoning-48471090eada https://pub.towardsai.net/the-air-gapped-chronicles-the-cour...
- tudorhn 6mo ago
  Author here. Would you dispute that they're much better than two years ago? I, personally, am quite impressed at how quickly they improved. It's not yet at the level for independent decision-making, but if the rate of improvement holds, I wouldn't be too surprised if we got there in the near future.
  - jqpabc123 6mo ago
    Would you dispute that they're much better than two years ago? I will readily admit that the guesswork has improved. Will you admit that guesswork remains the foundation of what they do? Guesswork will always be suspect. Removing it will require a fundamental redesign and may not be achievable within the constraints of current hardware and economics. My guess is any significant breakthrough will likely involve quantum computing and renewable energy.
    - tudorhn 6mo ago
      That's literally the thesis of the article. Even though guesswork remains the foundation, the improved training can help discover patterns, backtrack, etc. Basically, explore the problem space by "guessing" through it
      - jqpabc123 6mo ago
        Basically, explore the problem space by "guessing" through it This is where we disagree. "Guessing" (aka trial and error) is not a "legally" responsible way to make decisions in many cases.
        tudorhn 6mo ago
        Right, I think I understand. You are specifically talking about independent decision-making. I have been focusing on "problem-solving". Tbh, I have no opinion on the legal framework. I'm sure most decisions will be supervised, at least for a while.
        jqpabc123 6mo ago
        The technology is being marketed as providing "intelligence" --- not trial and error problem solving --- where "error" is a likely result. I wouldn't be surprised if the tool vendors aren't eventually forced to rename it and/or limit it's use in their terms of service due to legal liability. A good analogy is Tesla's "Full Self Driving" which was renamed to "Full Self Driving Supervised" --- an oxymoron.

4 points by tudorhn 6mo ago

Are they so good? They can excel at information retrieval. They often fail at logic and reason. Predicting the next word is basically guesswork. Not the sort of thing any responsible person should rely on to make important decisions. Doing so leads to liability issues in many business and professional environments. The legal system will have significant influence here. https://pub.towardsai.net/the-air-gapped-chronicles-the-court-asked-for-the-llms-reasoning-48471090eada https://pub.towardsai.net/the-air-gapped-chronicles-the-cour...

tudorhn 6mo ago
Author here. Would you dispute that they're much better than two years ago? I, personally, am quite impressed at how quickly they improved. It's not yet at the level for independent decision-making, but if the rate of improvement holds, I wouldn't be too surprised if we got there in the near future.
- jqpabc123 6mo ago
  Would you dispute that they're much better than two years ago? I will readily admit that the guesswork has improved. Will you admit that guesswork remains the foundation of what they do? Guesswork will always be suspect. Removing it will require a fundamental redesign and may not be achievable within the constraints of current hardware and economics. My guess is any significant breakthrough will likely involve quantum computing and renewable energy.
  - tudorhn 6mo ago
    That's literally the thesis of the article. Even though guesswork remains the foundation, the improved training can help discover patterns, backtrack, etc. Basically, explore the problem space by "guessing" through it
    - jqpabc123 6mo ago
      Basically, explore the problem space by "guessing" through it This is where we disagree. "Guessing" (aka trial and error) is not a "legally" responsible way to make decisions in many cases.
      - tudorhn 6mo ago
        Right, I think I understand. You are specifically talking about independent decision-making. I have been focusing on "problem-solving". Tbh, I have no opinion on the legal framework. I'm sure most decisions will be supervised, at least for a while.
        jqpabc123 6mo ago
        The technology is being marketed as providing "intelligence" --- not trial and error problem solving --- where "error" is a likely result. I wouldn't be surprised if the tool vendors aren't eventually forced to rename it and/or limit it's use in their terms of service due to legal liability. A good analogy is Tesla's "Full Self Driving" which was renamed to "Full Self Driving Supervised" --- an oxymoron.