I really like the reading of the LLM's responses you do in the early part of the essay. It nicely illustrates how the computational "thinking" an LLM does is quite different from the way a human thinks about using language. This difference is hard to get across, and you do a good job of it here.
Thank you for the comment, Rob. Very glad you found it interesting. It's hard to avoid anthropomorphizing, and of course I succumbed completely when comparing communicative competence, but I think it was useful to do so for illustrative purposes. I was considering adding a more elaborate evolutionary perspective with Kahneman's systems, but I am still thinking about this. A future paper, perhaps.
I really like the reading of the LLM's responses you do in the early part of the essay. It nicely illustrates how the computational "thinking" an LLM does is quite different from the way a human thinks about using language. This difference is hard to get across, and you do a good job of it here.
Thank you for the comment, Rob. Very glad you found it interesting. It's hard to avoid anthropomorphizing, and of course I succumbed completely when comparing communicative competence, but I think it was useful to do so for illustrative purposes. I was considering adding a more elaborate evolutionary perspective with Kahneman's systems, but I am still thinking about this. A future paper, perhaps.