Post #3986613

2026-07-19 18:15 UTC

Reminder: undisclosed slop machine usage, even in so-called "human in the loop" scenarios, is a form of reputation laundering. That is, you are using your reputation as a human with expertise to provide cover for a technology which stands in direct opposition to human expertise. If the excuse for not disclosing usage is "well people don't like it and will be mad at me," the appropriate question to resolve is why people don't like it, not why you unethically chose to hide it.

Replies (5)

@gary_alderson@infosec.exchange 2026-07-21 13:12

@jaredwhite@indieweb.social what about in 5-10 yrs - reputations blown apart - i don't buy it

Open ##3986612
@aurellence@todon.nl 2026-07-21 14:57

@jaredwhite@indieweb.social Yes! This! The human in the loop is the one who will take the fall when the slop machine inevitably screws up. They're scapegoats.

Open ##4028111
@synlogic4242@social.vivaldi.net 2026-07-21 15:25

@jaredwhite@indieweb.social yep the laundering happens both ways. the human might make the AIs results look better than by default, or vice versa. and makes it harder to gauge both the skill and LOE of the human in question. harder to determine if they are truly a fraud while at same time massively increasing chance they are. its a shitshow for signal conveyance to the rest of the community

Open ##4028115
@noodlemaz@mstdn.games 2026-07-22 06:42

@jaredwhite@indieweb.social I'd say the question is 'why did you use it then' rather than why people don't like it (although I suppose for the user, it is both)

Open ##4028116
@gabrielesvelto@mas.to 2026-07-22 07:38

@jaredwhite@indieweb.social one interesting wrinkle in this is that using LLMs without disclosing their use is poison for the training sets. Model collapse is directly caused by feeding model outputs into the training set, which is why the AI companies are now hell bent on having your work labelled as co-authored by a bot. It's crucial for them to be able to filter out those materials from their training set. In other words people using LLMs without disclosing their use are actively contributing to the downfall of the AI industry.

Open ##4028118