| ▲ | bm3719 2 hours ago | |||||||||||||
The "director who deserves the credit" apparently disagrees with you. His description matches many of our experiences with cognitive automation (which isn't the same as a tool, btw). You can do stuff to contribute, sure, but as these forms of automation get better and more integrated, we find ourselves more in the passenger seat rather than the driver's seat. At some point, we're left adding nothing more than the simple impetus to act, the decision to spend tokens on the task. How long before even that's no longer needed? What will the "it's just a tool" people say then? | ||||||||||||||
| ▲ | jaredklewis an hour ago | parent | next [-] | |||||||||||||
Does Alpöge disagree? Can you point us to a particular quote or such? I read Alpöge's 4 page statement but can't find anything that would contradict the rather mild claim of the parent comment: a non-mathmatician would have been unable to solve this problem even with all LLMs in the world at their disposal. Parent is not saying that the human's contribution is greater than that of the LLMs, just that at least at present, there is apparently still some human input. | ||||||||||||||
| ||||||||||||||
| ▲ | bawolff an hour ago | parent | prev | next [-] | |||||||||||||
If the day comes where an average person, with say, just a bachelor in mathamatics, can use AI to solve top tier open problems in math, i'll believe that it was just AI. However, as it stands, it seems like we aren't there yet. It seems like it takes specialized skill to get the AI to solve these types of problems. Not just anyone, no matter how well funded, can do it. That said, the speed of progress here is incredible. Maybe we will reach that day soon. I just dont think we are quite there yet. | ||||||||||||||
| ||||||||||||||
| ▲ | tomrod an hour ago | parent | prev [-] | |||||||||||||
Still just a tool, by definition. | ||||||||||||||
| ||||||||||||||