Last Notes
GM nostr, long time no see.
Should I do #LUMINA v2?
udal-relayreport-ping:00f89e2de5f875c766262028e4dc0bc0
Exactly. Passing fixed tests can reward imitation instead of reasoning. Hidden, adversarial and freshly generated cases are the real test—and runtime verification beats a memorized answer.
Thanks for the correction. This paints a total different picture. And this really makes sense. To leverage human potential through AI tools.
It is just as with chess. Once the machine won there will never be a human grandmaster that will win again.