In the field of statistics, there is risk from extrapolation beyond the range of the data; the risk is managed by changing assumptions and retesting. In the field of law, there is a risk of an argument that assumes facts not in evidence; the risk is managed by rigorous cross-examination.
Fake legal research, while sensational, is a clear misuse of the tool. That's not what I did. The 26 pages took 26 pages because of my best efforts to test, retest and cross-examine the responses. Latin is clearly a topic on which there is substantial information in the underlying dataset, still there were definitely errors: some I caught, some it caught itself, some not caught, some you and the others here would have caught.
The AI had (by definition) the patience to both translate and explain (and re-explain, and re-explain again) every Latin suggestion. This helped me. I found it interesting that it ended up suggesting phrases you suggested, and when fully explained, I got onboard.
ChatGPT is not a Latin scholar, nor is it a 3-year-old. It is also not Google Translate. It can deal with context, make fine-grained adjustments in response to careful questioning, and is accessible in a way that people often are not. At some point, a general tool of very high functionality will be more useful than a very specific tool that is hard to master.
An analogy might be a manual versus automatic transmission in a car. With a skilled operator, manual transmissions were for many years more fuel efficient, faster, etc. Today, the opposite is true. Yet, as a qualitative matter, I have zero interest in owning a car with an automatic transmission. At some point, when ICE is regulated away, I will have no choice but to drive an automatic EV. That will suck for me, while for nearly everyone else driving they will be fine with it, not knowing or caring about the lost art of heel-and-toe.
Clearly these matters are distasteful to you. It's your prerogative to ban, edify, scupper the bilge, or whatever. As Ken Kesey once said, though: If you've got it all together, what's that all around it?