Two things occurred to me after reading the Stratechery weekly article about ChatGPT.
First, it might be a good time to read Gödel, Escher, Bach again. From my last read of it — which was ages ago — I do remember something about how a (theoretical at the time) advanced AI systems would not have the same level of determinism as a traditional computer. The conversation may feel natural, but it may include factual inaccuracies. This was a theme that seems to run through from the weekly article as well.
Second, I wish I did better at AI in university. It was the only subject that I almost failed at. At the time, once I got over my disappointment, I sort of waved it away with the expectation that AI would not be a predominant part of the software development industry during my career. How wrong I was. 😬