Thread

1. js8 · 1790928479 · focus · HN ↗

   I am also an optimist about AI, although I worry about the labor aspect as well (as a leftist I see the class aspect of this, on the other hand, as a computer programmer, computers were also originally meant to replace human labor, and we all see what happened).

   My view is, current top AIs do multiple different things:

   1. Interpret and output natural language

   2. Do formal logical reasoning

   3. Do informal logical reasoning

   4. Provide encyclopedic knowledge

   5. Discern vague instructions/statements and fill the most likely gaps (kind of error correction)

   6. Serve as a model of the human mind, a proxy for a person

   7. Translate between different languages, formal and informal

   All these things are combined in these huge, hard-to-understand blobs of weights. I think humanity would be better off by understanding each aspect separately, but disentangling them will take decades of philosophical research. So there is a lot of interesting work ahead of us.

   1. pwndByDeath · 1791039083 · focus · HN ↗

      Doing formal logic because they memorized the token patterns of humans doing formal logic.

      1. saulpw · 1791042550 · focus · HN ↗

         Isn't it remarkable that this works? I wouldn't have thought so either, and yet here we are.

         1. pwndByDeath · 1791047572 · focus · HN ↗

            Not really, I think it says more about how little intelligence is in the behaviors we value
      2. whattheheckheck · 1791043675 · focus · HN ↗

         You can ask it to provide a prolog or datalog program

         1. pwndByDeath · 1791054784 · focus · HN ↗

            The inherent 1st order thinking in that way of thinking... It highlights just how little we understand what we understand, and how utterly ignorant we are of what we don't.

            I'm already seeing new "engineers" who cannot work things out without AI holding their hand

            1. Valodim · 1791055620 · focus · HN ↗

               Like those people who can't drive stick shift. What do they even do if they need to operate a real car?

               1. pwndByDeath · 1791082013 · focus · HN ↗

                  Perhaps I'm not intelligent enough to understand, Many people don't exercise thinking, many educated and capable people prefer not to, comparing that state to LMs and calling it dangerous intelligence is a bit too much emphasis on the later even of the first part may fit. It could be dangerous, like letting someone who can't drive stick loose in traffic.
      3. js8 · 1791139850 · focus · HN ↗

         Yes, but LLMs are doing other things than just what Prolog does, and that's my point. People don't get excited about Prolog helping them write software, even though it has some capability to do so. LLMs contain other potentially useful algorithms (of not just logic, but epistemology in general, it understands how to create hypotheses and run experiments to confirm or refute them).

         Also, of course, any computer program can be considered a logical calculation when interpreted in lambda calculus. And LLMs are computer programs, so you can consider them a formalization of all the above in the formal logic (although not a very good formalization, for various reasons).

         1. pwndByDeath · 1791175934 · focus · HN ↗

            I guess the point I'm trying to make is though the outputs are the same, the LMs not "doing" formal logic, its not in it, it's doing something else that only fools us because many many others have done formal logic in the training set

Unofficial Hacker News client; not affiliated with Y Combinator.