More making ChatGPT look silly
2024-06-09 16:41:30.372205+02 by Dan Lyke 0 comments
The thing about reading Demian's Blog: LLMs are not even good wordcels is that I got to
In typical ChatGPT fashion, it spouts a lengthy answer that makes it seem like it’s reasoning about which letters are missing from the previous pangram attempt.
And I had reiterated that feeling that maybe the positive reactions to LLMs are from people whose educational process had a lot of reward for "rationalize about this answer" rather than for correct answers, and the wordy bullshit generation fills that role nicely.