General Discussion

And struggling to find citations, many of the citations were made up or at least didn't seem to exist via regular searching. Obviously none of them were used in my work, nor did I use the AI to write anything. But I think that's very much the nature of LLM's in that it's at least partially a fancy text predictor. So for things like coding, where there's generally a preferred way to do something, it works to provide a reasonably close output. But they suck at saying no results found or telling you how confident the answer is, so everything is spewed out the same.

But even for coding which is a best case for an LLM, you're still shortcutting your knowledge process which includes failing, which includes finding things you didn't know that might work for another problem or another implementation. Sure in theory it might foster interest in learning more in some people, but most people will take the easier short term path of completing the assignment.