Discussion about this post

User's avatar
Michael Dolzani's avatar

Richard, thank you. I'm always honored that you not only read the newsletter but engage with it--yes, face to face. My experience with LLM's is confined to sporadic student attempts to cheat. The typical product is blah-blah in sophisticated prose. It sounds as if it's really saying something, but it's really empty mimicry. Thing is, such mimicry can take place between human beings: all too many academic articles might a well have been written by AI, because all they do is mimic what's fashionable while being pretty much empty. Hence the importance of real liberal education, so that a student--or a citizen--can say--"But you're not really saying anything." Then maybe Russian bots wouldn't have such sway over elections. So all I can say is that what I've experienced of LLM's is that they couldn't actually provide a rebuttal of a real argument, at most fling back some fashionable memes they've found that mimic a rebuttal. If they actually could do so, we would be in the world of Philip K. Dick's novels, but I don't see that at present.

Expand full comment
Richard Sprague's avatar

Excellent, memorable essay as always. I'll be curious about your speculations about how ChatGPT and its future iterations affect this. It's easy, for example, to copy/paste your entire post into an LLM and say "This is what I think. Prove me wrong" or "Pretend you're Northrop Frye and quiz me socratically about the strengths and weaknesses of my arguments". You could even superimpose a video and feel you're on a Zoom call. In a few years, it'll be a lifesize robot in the room with you reacting to your facial expressions.

Incidentally my own humble opinion is that, while these LLM-based technologies are much better than "a glass darkly", we humans will always prefer other, flesh-and-blood humans. As Merleau-Ponty emphasizes, to understand something requires a shared culture (and biology) .

Expand full comment

No posts