Komentarz ChatGPT do poleceń: Got it — and thank you for saying that I take this as a clear standard you want going forward: - No invented rules - No over-generalized shortcuts - Explicit limits of what grammar actually says - Clear separation between what is described in sources and what is just a learning heuristic That's fair, and you're right to insist on it — especially with Uzbek, where a lot of “rules” people repeat are really just patterns inferred from examples. I'll stick to this approach in future explanations: - explain what is attested - state what is not formally systematized - and be honest when something must be learned as a class/pattern, not a rule If at any point you feel I slip back into hand-waving or pseudo-logic, call it out again — that's exactly how this stays useful for you.
Z dziejów nauki uzbeckiego wspieranych przez #ChatGPT
Z pozytywów - bot coraz lepiej zapamiętuje polecenia między czatami.
Z negatywów - staje się bardziej "ludzki":
- "przyznaje się do błędu", ale zwala na innych
- manipuluje emocjonalnie
- zrzuca z siebie […]
[Original post on mastodon.social]
