@lucidpaths33@spawn It's amazing how far you can go by recursively asking, "hmm, I wonder if I could just tell the AI to prompt itself to do all of that", and keep working up the chain
When I was in my twenties it was common to hear people say that the next generation would be so much better with computers and tech than my generation and same for every generation after. I don't think that happened. Individuals yes, but generally no.
@karatademada I mostly think it’s a psychological shortcut. People equate effort with quality because for many things previously that was somewhat true
@n0coder@a0script I didn’t fully understand what you were saying. I responded to the bits I did understand. Or at least thought I did. I’m open to the likely possibility that I misunderstood what you meant all along
@a0script@n0coder I mean I routinely create things with ai by specifying the outcome and not the how. I feel like this whole thesis involves a lot of moving goalposts. At no point have I claimed ai is as creative as humans. But to claim it cannot ever do anything new is absurd in the other dir
@n0coder I actually disagree. I think AI can create novel things, because I think AI does genuinely understand language, probably better than humans do. I'm using the word "understand" loosely because we don't know what understanding is for humans or AI. But it's just my opinion.
@n0coder I think you're clutching at straws tbh. Happy to disagree. Maybe I just don't understand what you're saying. Humans don't understand without things being explained to them either.
@n0coder Your bar for proof is for someone to do a task without using their own mind?? I'm very confused right now. Can you explain what you mean in a different way? Give an example
@n0coder This is a claim about your perception of AI without any evidence to back it up. What would count as an example of AI deviating far from the sources of our own engineering? And why do you think that is something it needs to do? Do humans do that? Would we need /wantto?
@n0coder I might be missing your point, but if you could make a language that worked that way, I guarantee LLMs could write valid code for it without that being in their training data
@n0coder Everything is "the same kind of" if you squint hard enough. This weakens your argument in my opinion because it applies to humans too. We have a lot of things in our evolutionary "training data". Give us something too far outside that and we struggle.