News for hackers. Stuff that matters.

A comment on LLMs do not merely reflect the bias of their training, they police it.

Re:LLMs do not merely reflect the bias of their training, th... by GL26 on Monday June 22, 2026 @07:23AM (#48628704)
We are not yet at misalignment, but this shows the existence of a slope that derivates into misaligned adversarial ai models. Must this be fixed at training time (at which step ?) ? Thinking about this report : https://ai-2027.com/
[ Parent | on HN ]
Re:LLMs do not merely reflect the bias of their training, th... by rwmj on Monday June 22, 2026 @07:44AM (#48628863)
That was a nonsensical work of fiction, not a report.
[ Parent | on HN ]