One of the most interesting results of LLMs having "gotten too good" is not that they can make convincing sounded errors but that we are often so resolute on th

One of the most interesting results of LLMs having "gotten too good" is not that they can make convincing sounded errors but that we are often so resolute on the idea we will be able to tell if something is AI generated based on how good we thought it was. The line has been in the edit history for at least 13 years but I wonder what happens to our conclusions when content is expected to have been created after "decent AI" was around https://en.wikipedia.org/w/index.php?title=Effects_of_stress_on_memory&oldid=431858437#:~:text=These%20stress%20hormones%20are%20also%20hindering%20the%20hippocampus%20from%20receiving%20enough%20energy%20by%20diverting%20glucose%20levels%20to%20surrounding%20muscles https://en.wikipedia.org/w/index.php?title=Effects_of_stress....