Sure, but it is a token predictor, and it is not intelligent not because it works by predicting a token, but it shows dumb behavior...
LLMs are a great search tool. It searches connections in the collective human knowledge that humans have written down through all the years....
They are very good at it, and that is about it.
Putting LLMs in loops with tool calls has created tons of emergent behaviour beyond simply being a search tool
If you search randomness, you ll get any behavior that you can imagine. That does not mean that the tool is NOT doing a search.
In the case of LLMs, it is not searching the whole of the randomness, but instead it just search among the grammatically correct sentences that is consistent with the existing patterns found in the existing written down human knowledge.
The thing is searching an associative memory for similar solutions and predicting what will work is not that far off conceptually from what humans do
I don’t think LLMs will reach the same level as humans but I would argue it is still a form of limited intelligence
It’s not a binary thing
It is not Intelligence. It is pattern matching.
https://news.ycombinator.com/item?id=49694427
Do you have any examples of this?