[Everyone disliked that]
https://www.youtube.com/watch?v=WK89SGSEf1s&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20251216-amazon-discovers-ai-sucks-at-automated-tv-recaps - podcast
time: 4 min 46 sec
That’s called collecting empirical evidence, I can respect that, hahahahahah.
What puzzles me is, did anyone watch it before publishing? Who approves these things?
Referencing the telephone game does not prove anything here. The telephone game is shows that humans are not good at copying something exactly without changes, which computers are better at. But the question here is if AI can achieve deeper understanding of a work, which is needed to produce a good summary. This is something humans are far better at. The AI screws up the summary here in ways that no reasonable person who has watched the TV series (or played the games) would ever screw up.




