Post #3940700
2026-07-19 10:09 UTC
@mikalai@privacysafe.social @troed@swecyb.com @david_chisnall@infosec.exchange Right, and they are pretty good at guessing by now …
Replies (1)
-
@mikalai@privacysafe.social 2026-07-19 19:30
@hfalcke@mastodon.social Unfortunately, guesses are picked up as gospels, or more or less good advice from human pro. But it is a guess without modicum of check with reality. Do you know why all of a sudden these agentic things are good with coding, like json output? Injected parsers. Read injected reality checks. But. Without nuances of implementation public is sold that there is an improvement of an underlying LLM guesser. The smoke is the problem. @troed@swecyb.com @david_chisnall@infosec.exchange