@nlnetlabs@social.nlnetlabs.nl
2026-09-16 05:58 UTC
Replies (6)
-
@happyborg@fosstodon.org 2026-09-16 10:02
@nlnetlabs@social.nlnetlabs.nl this is good, but any normalisation or encouragement for people to use LLMs is at the least questionable because of the widespread and serious harmful effects all use has. While distracting people with claims they're building a doomsday machine, the real, immediate harms are not being given attention, so I urge you and others to reconsider accepting any form of input based on use of #LLM tech.
-
@rgacogne@mamot.fr 2026-09-16 10:05
@nlnetlabs@social.nlnetlabs.nl Wonderful post! Many of us are in the same boat.
-
@zelf@sunbeam.city 2026-09-16 15:13
@nlnetlabs@social.nlnetlabs.nl Great to see you take this stance! It's been an active discussion in local tech-coop scene and having y'all lead the way also paves the way for others ⭐️
-
@nyquildotorg@gts.nyquil.org 2026-09-16 19:26
@nlnetlabs@social.nlnetlabs.nl I think it may have been a mistake posting this on a network full of neurodivergent people who will happily argue that even though your conclusion is the correct one, your rationale doesn't make any damned sense, therefore you are wrong.
-
@stefan@boeck.social 2026-09-16 23:08
@nlnetlabs@social.nlnetlabs.nl The line that carries the post for me: the contributor cannot answer for the design. "Authored by a human" is the proxy a policy can state, but nothing a maintainer can check. What can be checked is a record: which prompt led to which file change, under which model, who signed it off. Then "I dunno, it's what Claude came up with" shows before review, not after. I build with an assistant daily and would want that record on my own contributions. Not as a defence, as a receipt.
-
@alexband@hachyderm.io 2026-09-16 08:52
@barthilhorst@mastodon.social ooooh nice catch! I was too focused on the writing to pick up on that. 🔥