I would say it’s a good example of a bad use-case for an LLM; you don’t have sources, and you can’t fact-check anything. Those two are absolutely vital requirements when claiming something as true.

Aside from that, most generative AI have been trained on vast amounts of data that was never allowed to go into the dataset; copyrighted/IP-righted paintings, articles, comics, and novels have been included against the wishes of the artists/authors. The fact that nothing is being done on a legal system level shows that copyright and IP rights clearly do not apply to American oligarchs, and many of us don’t like that. Most generative AIs also need an absurd amount of power to run and hurt the environment a lot. It sucks to separate paper and plastic waste just to know that there are people blasting through an hour worth of airconditioning just to ask a computer something they could’ve looked up instead (and found sources, too!)

I say this as someone who loves using AI and experimenting with it^1^: This was a very bad use-case of generative AI.

^1^: although lighter ones and locally like Mistral, using open datasets like OpenOrca