And now, for something completely different, there's this elucidating yet not really surprising tidbit (courtesy of Wired):
A new study from the University of Texas at Austin, Texas A&M, and Purdue University shows that large language models fed a diet of popular but low-quality social media...