22 ms·
I feel like this is how normal work is. When I have to figure out how to use a new app/api etc, I go through an initial period where I am just clicking around,
by whinvik 1y ago
I feel like this is how normal work is. When I have to figure out how to use a new app/api etc, I go through an initial period where I am just clicking around, shouting in the ether etc until I get the hang of it.
And then the third or fourth time its automatic. Its weird but sometimes I feel like the best way to make agents work is to metathink about how I myself work.
- suchintan 1y agoI have a 2yo and it's been surreal watching her learn the world. It deeply resembles how LLMs learn and think. Crazy
- melagonster 1y agoI am sorry, but you are scoffing at the humanity of your kid; you know that, right?
- Retric 1y agoOdd, I've been stuck by how different LLMs and kids learn the world. You don’t get that whole uncanny valley disconnect do you?
- goatlover 1y agoHow so? Your kid has a body that interacts with the physical world. An LLM is trained on terabytes of text, then modified by human feedback and rules to be a useful chatbot for all sorts of tasks. I don't see the similarity.
- crazygringo 1y agoIf you watch how agents attempt a task, fail, try to figure out what went wrong, try again, repeat a couple more times, then finally succeed -- you don't see the similarity?
- dingnuts 1y agono I see something resembling gradient descent which is fine but it's hardly a child
- balder1991 1y agoNo, because an agent doesn’t learn, it’s just continuing a story. A kid will learn from the experience and at the end will be a different person.
- CaptainOfCoit 1y agoYou just haven't added the right tool together with the right system/developer prompt. Add a `add_memory` and `list_memory` (or automatically inject the right memories for the right prompts/LLM responses) and you have something that can learn. You can also take it a step further and add automatic fine-tuning once you start gathering a ton of data, which will rewire the model somewhat.
- haskellshill 1y agoPerhaps it can improve but it can't learn because that requires thought. Would you say that a PID regulator can "learn"?
- CaptainOfCoit 1y agoI guess it depends on what you understand "learn" to mean. But in my mind, if I tell the LLM to do something, and it did it wrong, then I ask it to fix it, and if in the future I ask the same thing and it avoids the mistake it did first, then I'd say it had learned to avoid that same pitfall, although I know very well it hasn't "learned" like a human would, I just added it to the right place, but for all intents and purposes, it "learned" how to avoid the same mistake.
- haskellshill 1y agoThis is a silly definition of learning, and any way, LLMs can't even do what you describe.
- deadbabe 1y agoA person is not their body. The person is the data that they have ingested and trained on through the senses that are exposed by their body. Body is just an interface to reality.
- haskellshill 1y agoThat is a very weird and fringe definition of what a person is.
- deadbabe 1y agoIf you have a different life experience than what you had so far, wouldn’t you be a different person?
- haskellshill 1y agoIf you had one fewer leg, wouldn't you be a different person? Implication is not equivalence.
- haskellshill 1y ago> It deeply resembles how LLMs learn and think What? LLMs don't think nor learn in the sense humans do. They have absolutely no resemblance to a human being. This must be the most ridiculous statement I've read this year