Raw LLM Responses

>...the parrot repeats these sounds until it is capable of a near-perfect mimicry... [Parrots are pretty frickin' smart](https://www.nationalgeographic.com/animals/article/year-of-the-bird-brains-intelligence-smarts). They're more than just mimics. Seems like an odd choice for leading in to this argument. On the one hand, sure, ChatGPT is doing a kind of mimicry. A very elaborate kind of pattern detection and pattern matching. I wouldn't suggest it has interiority. And yet. "Intelligence" has been notoriously hard to define in a rigorous way. And today it looks like the things ChatGPT can do cover a *disturbing* fraction of what we mean when we say "intelligence." Maybe what ChatGPT does isn't *sufficient* to count as "intelligent," but its capabilities are probably *necessary* to count as "intelligent." >If I asked ChatGPT to write a review of Star Wars Episode IV, A New Hope, it will not ... appreciate how the script, while being a trope-filled pastiche of 1930s pulp cinema serials, is so finely tuned to deliver its story with so few extraneous asides... A human who is less than 40-50 years old probably wouldn't do that either, though. Star Wars may have been a formative experience for you (and for me) but it wasn't for younger people. You might say it wasn't a prominent part of their training set. On the other hand, if you by chance respond that you're under 40 and yet you have a deep respect and appreciation for Star Wars that is different from what an AI would have? Well. Wouldn't that largely be down to all the criticism and appreciation of it that you've absorbed (just as an AI would), not because you experienced it in its full cultural context yourself? >Crucially, if all of the source material is bunk, the output will be bunk. Also true of human intelligence. (How many people "believe in" QAnon?) I don't think that's a crucial distinction; I don't see how that's any sort of distinction at all. >What I think is happening, here, when peop