I tried feeding ChatGPT a misleading prompt. It called me out on my mistake (which is good) but then told me a deeply wrong story about what it thinks WAS correct.
No, Paul Upchurch never synonymised Apatosaurus with Diplodocus. But neither did Riggs, and his 1903 paper on Brachiosaurus never even mentions either of those genera. (His 1904 paper on Apatosaurus synonymises Brontosaurus with it, but not Diplodocus.)
This thing is a lie generator.
@mike ChatGPT is pretty good on subjects that a lot of techies know about—not a surprise—which means it's lousy on paleontology. Probably even worse on, say, art history. Calling the training corpus "skewed" is maybe the most polite way to put it. 😀
@medigoth That's a good point. I might try it on some computer science stuff next time around.
