I wrote about an exchange I had with some of my computer-oriented friends yesterday and why ChatGPT is bad for academic papers, even if you're just using it to find sources. joshuapnudell.com/2024/10/04/s...
Yesterday I ended up in a long conversation with friends sparked by this article. They are computer people (programmers, IT, web admin types) and largely pro-AI, while I, the historian and teacher,…
I basically by default skip the AI search result on Google because it is so unreliable.
This reminds me of the lawyers who used ChatGPT to find legal precedents that supported their client's case and were provided with fake judicial opinions complete with fake quotes and citations and companies referenced that never existed.
Thank you for articulating this so well, I instinctively felt the same, but have struggled to identify exactly what the problem is. I have seen a deterioration in the quality of student essays over the past year, and wondered if this might be the reason.
A student asked me about this the other day, and this is a really clear example of the problem! Thank you.
there's an einstellung effect problem too: the more anyone uses the plagiarism machine, the more they rely on using it. they don't learn what they're using chatgpt for, they're learning to use chatgpt. and when chatgpt is incorrect, the person is eventually unable to spot what's wrong in the output
A friend shared a post from an FB page yesterday which had entirely made-up citations. He then asked ChatGPT to list his own top 10 publications, and the result included not a single paper that he'd actually written (none of the papers existed). I would NEVER trust it to find me sources!
Can anyone point me to a critical bibliography on AIs? I am hearing more of this glassy-eyed boosterism where I work and it would be helpful if I could give people a list of sources.
I tell writers: a human being in front of a blank page is like a cat in front of a mouse, an osprey hovering over a river The blank page fears *you*