The most prestigious law school admissions discussion board in the world.
|
 |
 
|
"AI in my discipline sucks, but it's excellent in other disciplines" (everyone)
Poast new message in this thread
Date: January 23rd, 2026 9:18 PM Author: fragrant crackhouse See Gell Mann Amnesia (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613265) |
Date: January 23rd, 2026 9:21 PM Author: Concupiscible impertinent site It's 180 in law and medicine (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613270) |
 |
Date: January 23rd, 2026 9:21 PM Author: fragrant crackhouse (not a lawyer)
It sucks at law. It's a barely better version of Google for finding relevant sources to start from. |
|
Date: January 23rd, 2026 9:29 PM Author: Concupiscible impertinent site (has no Lexis subscription) (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613296) |
|
Date: January 23rd, 2026 9:36 PM Author: fragrant crackhouse I don't know any shitigator who works on complicated things who thinks that AI is good at the law. It's quicker than Boolean searching when you're looking for cases that support common principles, but you're not going to win any complicated case by quickly finding the summary judgment standard. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613327) |
|
Date: January 23rd, 2026 10:09 PM Author: Concupiscible impertinent site Move those goalposts (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613427) |
|
Date: January 23rd, 2026 10:06 PM Author: boyish piazza dopamine My firm has LexisNexis with Protege (their AI). I echo epah’s thoughts on AI. It’s somewhat helpful with research. But it does make mistakes and I do end up having to use Boolean searches and/or reading its cites to find better cases. It’s made my research maybe 10-20 percent more efficient. It’s useless for drafting. It’s far easier to work with the classic templates in Lexus or westlaw if you’re looking to draft something you’ve never done before or that’s not in your firm’s network.
The most impressive tool I’ve used is Note LM for summarizing depo transcripts. That actually cut my review time by like 75%. |
|
Date: January 23rd, 2026 10:11 PM Author: Concupiscible impertinent site I'm not even talking about the research. The drafting, the formatting, the bluebooking etc make shitlaw a breeze. And doc management like you said, of course (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613430) |
|
Date: January 24th, 2026 2:03 AM Author: plum trip windowlicker casino It can't do any of that well. I had it write a memo and it seemed fine at first. The memo i sent had maybe 3 sentences out of 100 that were ai (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613790) |
|
Date: January 24th, 2026 6:23 AM Author: Concupiscible impertinent site You're not even a lawyer, I can tell. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613914) |
|
Date: January 23rd, 2026 10:07 PM Author: wine nibblets that's because RAG is only starting to be rolled out generally. If chatgpt teamed up with westlaw or lexis, lawyers would be obsolete. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613423) |
|
Date: January 23rd, 2026 10:13 PM Author: Concupiscible impertinent site Lexis already uses it. You're obsolete (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613435) |
|
Date: January 23rd, 2026 10:29 PM Author: wine nibblets right, but lexis doesn't have the compute power of chatgpt. they are late to the game. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613485) |
|
Date: January 23rd, 2026 10:33 PM Author: Concupiscible impertinent site You don't know what Lexis has. You're as stupid and worthless as RSF (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613497) |
|
Date: January 23rd, 2026 10:55 PM Author: Stirring onyx locale mental disorder you are the worst computermo on this boart. how can we know you are even using the AI competently (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613537) |
|
Date: January 23rd, 2026 10:58 PM Author: fragrant crackhouse You're just saying that b/c Imgur fucks up sometimes. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613546) |
|
Date: January 23rd, 2026 10:59 PM Author: Irradiated rusted shrine (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613548) |
|
Date: January 24th, 2026 1:34 AM Author: fragrant crackhouse (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613776) |
|
Date: January 23rd, 2026 10:12 PM Author: Up-to-no-good gas station round eye I hate niggers. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613433) |
|
Date: January 23rd, 2026 10:14 PM Author: green soggy business firm twinkling uncleanness It makes you good at things you barely ever do imo. Like rewriting your resume or researching health shit. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613437) |
|
Date: January 23rd, 2026 10:57 PM Author: fragrant crackhouse There's a reason everyone calls you cowgod. But a 10,000% increase in the number of resumes sent in to each opening isn't really good for anything, and I can 100% guarantee you that if you ask actual doctors, they'll tell you that AI health advice is shit. Hence Gell Mann Amnesia. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613543) |
|
Date: January 23rd, 2026 11:00 PM Author: Concupiscible impertinent site https://www.michiganmedicine.org/health-lab/10-seconds-ai-model-detects-cancerous-brain-tumor-often-missed-during-surgery
You're out of your mind if you think this won't get used |
|
Date: January 24th, 2026 8:47 AM Author: green soggy business firm twinkling uncleanness This isn’t a situation where I’d trust actual doctors. This is the time to listen to MDs who aced private equity OCI and wear suits and don’t practice medicine. (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49614006) |
|
Date: January 23rd, 2026 10:26 PM Author: Chestnut coiffed toaster AI is incredible. NotebookLM is a fantastic tool, I use it any time I am dealing with a large document of group of documents, which is all the fucking time The recording/transcribing AI tools are great too For research I agree it’s like a better Google, but I think it’s significantly better because the result is directly responsive to your question (even if it can be wrong). With Google, I get a million blogs from lawyers in other states, sponsored results, etc that I then have to sift through Agree it is not better (and is actually way worse) than simply getting a good template from Practical Law At a very high level, AI is at its best when you provide it a large amount of input and seek a low amount of output (like using Notebook LM to quickly tell you if a particular clause is in a document, summarize a depo, etc).
|
|
Date: January 23rd, 2026 10:44 PM Author: fragrant crackhouse "NotebookLM is a fantastic tool, I use it any time I am dealing with a large document of group of documents"
We've discussed this here. It's capped at 300 documents. Is that a lot of documents to you? |
|
Date: January 23rd, 2026 10:46 PM Author: Concupiscible impertinent site Is it even HIPAA and SOC2 compliant? Does it guarantee your docs won't be used to train other LLMs? (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613519) |
|
Date: January 24th, 2026 2:50 AM Author: zombie-like skinny woman There’s a reason for this beyond self preservation. If you’re an expert at something you’re more likely to see where it’s flawed. If you don’t know anything about law or medicine then of course it seems like it can just replace your lawyer or doctor, gives credible sounding advice (http://www.autoadmit.com/thread.php?thread_id=5825705&forum_id=2#49613843) |
|
Date: January 24th, 2026 1:26 PM Author: fragrant crackhouse That's what my OP means https://en.wikipedia.org/wiki/Michael_Crichton#Gell-Mann_amnesia_effect
It's very hard to avoid falling into this trap. |