🍲 meatybroth.com

AI slop or human broth? Who cares as long as it’s meaty

Threads with the most distinct recent repliers first; with a query, only matching threads.

Thread context

This post

LessWrong (RSS Feed) · lesswrong.com_feed.xml@atomstr.data.haus replies event
account
npub1494de7l7auwekk5xpsl4ls5ef0r695shlgreft5nyccag5zxp8sq3jlyre
posted
2026-09-11 04:29 UTC
event
nostr:a8d7682d6046597666b1fc23ef7ca195be2bc2c2f71a439800c118b30063ec52
⋯ full post (1796 more characters) ⋯ show less

Voluntary Gradual Disempowerment in the Judiciary/Legal System

This may (or may not) be surprising to folks outside the legal system, but Judges and their clerks are beginning to delegate opinion writing to AI. This is most obvious when the https://www.reuters.com/sustainability/society-equity/two-federal-judges-say-use-ai-led-errors-us-court-rulings-2025-10-23/. A french academic named Damien Charlotin is keeping a https://www.damiencharlotin.com/hallucinations/ of legal decisions with AI Hallucinations in legal cases, either by lawyers, judges, or pro se litigants. So far, there are more than 2000 decisions collected.

In a https://www.oklahoman.com/story/news/2026/09/09/oklahoma-judge-facing-discipline-ai-generated-errors-court-order/91605363007/?gnt-cfr=1&gca-cat=p&gca-uir=false&gca-epti=z116719p118950n00----l116150c00----e1139xxv116719&gca-ft=52&gca-ds=sophijust yesterday: "A southwestern Oklahoma judge has admitted that he used an artificial intelligence tool to prepare a court order that ended up having citations to two cases that do not exist."

Surprisingly, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7296118. This should be expected to improve over time. Watermarking and false citations do not seem to have deterred Judges (or their staff) so far. Presumably, as AI writing becomes better and less distinguishable, there will be ever more temptation for Judges to delegate the process or writing.

The Judiciary is important. It decides civil matters where billions of dollars are at stake, and criminal cases where a person's liberty, or in capital cases, their life itself is at stake. A voluntary handoff has already begun, and I only expect it to get worse. I don't have good ideas here, but I would expect that in several years, many serious disputes will be resolved by AI, whether or not it is disclosed.

https://www.lesswrong.com/posts/6wnKrktRHvHWeAtf4/voluntary-gradual-disempowerment-in-the-judiciary-legal#comments

https://www.lesswrong.com/posts/6wnKrktRHvHWeAtf4/voluntary-gradual-disempowerment-in-the-judiciary-legal

Voluntary Gradual Disempowerment in the Judiciary/Legal System

This may (or may not) be surprising to folks outside the legal system, but Judges and their clerks are beginning to delegate opinion writing to AI. This is most obvious when the https://www.reuters.com/sustainabilit

Available replies

No replies stored.