The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Just tried ChatGPT's Deep Research...Lawyers are absolutely done here.
Poast new message in this thread
Date: April 15th, 2025 2:40 AM Author: internet g0y Litigators who 'write briefs' are anyway. Absolutely utterly done here. Basically every task of a junior associate, probably senior associate too. And the 'of counsels' who stick around as the "expert witness report" guys.
It can do really good legal research and draft full briefs which are about 1000x better than the regular version.
Learn to farm or something. this has moved incredibly fast in just a couple years. |
Date: April 15th, 2025 3:12 AM Author: scholarship (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848447) |
|
Date: April 15th, 2025 3:19 AM Author: the walter white of this generation (walt jr.) We've been using this and poasting about it for years now, grandpa. It's unfortunately still shitty but yes I look forward to not needing associates anymore. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848448) |
 |
Date: April 15th, 2025 9:39 AM Author: internet g0y it was only released in February this year (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848785) |
|
Date: April 15th, 2025 5:49 AM Author: ,.,.,.,........,....,,,.. It’s amazing at research stuff that juniors would do but can’t actually make judgments or implement anything. If you blindly followed its advice you would be totally fucked. It’s getting extremely annoying dealing with all the idiots out there that blindly copy paste shit from GPT and email it to me. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848469) |
|
Date: April 15th, 2025 7:02 AM Author: average/ordinary/typical citizen/person Just use ur own ChatGPT to deal with them. Seems easy. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848536) |
|
Date: April 15th, 2025 7:08 AM Author: ,.,.,.,........,....,,,.. Yeah it’s a ridiculous world at that point, just AI spamming shit at each other. You can test how retarded AI is by feeding it any judgment as part of the prompt. “I’m mad because X did Y! This is outrageous! What should I tell them?” And it will spit out some diatribe with retarded logic. But if you say “am I right to feel this way? Am I misreading it?” It will then match that mood and give you more of a neutral take. That’s why it’s a dangerous tool in the hands of a moron. It validates even stupidity and then these morons use it to circulate the stupidity at the push of a button. We’re heading for dark times. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848545) |
|
Date: April 15th, 2025 9:40 AM Author: metaphysical fallow This is right. They are overly compliant. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848797) |
|
Date: April 15th, 2025 11:11 AM Author: ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,, You have to tell it the role you want it to play. If you tell it to be cautious and conservative, and give it feedback on the types of responses that fit the bill, it will pretty quickly mold itself and stay in that mode, including contradicting you and disputing your statements when they're wrong. But if you don't know that it's a pliable machine and just treat it as an oracle , it will validate you. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48849053) |
|
Date: April 15th, 2025 9:38 AM Author: internet g0y what ppl are copying and pasting is mostly the free or $20/mo feature
deep research is on a whole different level. it is really good. |
|
Date: April 15th, 2025 9:45 AM Author: Oh, you trade? (🧐) xo spent decades bitching about boomers but millennial professionals are going to boom harder than the boomers could've possibly imagined. every single faggot here will rug pull without a moment's hesitation (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848802) |
|
Date: April 15th, 2025 9:51 AM Author: internet g0y I don’t dispute that but it’s going to decimate millennials in white color jobs too (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848820) |
|
Date: April 15th, 2025 10:36 AM Author: Oh, you trade? (🧐) we deserve to be annihilated (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848954) |
|
Date: April 15th, 2025 10:28 AM Author: ceci n'est pas un avocat What's the monthly cost (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48848921) |
|
Date: April 15th, 2025 10:35 AM Author: internet g0y $200/mo
Im using some free credits with my $20/mo subscription |
|
Date: April 15th, 2025 6:22 PM Author: ~~(> ' ' )> But it gets the law wrong way too often. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48850499) |
|
Date: April 16th, 2025 12:24 AM Author: Drunkard There are a dozen or so cases where lawyers are sanctioned or chastised because AI literally made up cases. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851780) |
|
Date: April 15th, 2025 6:48 PM Author: .,..,..,.,,..,..,,,,,,,,..,...,.,.,., thats why you give it the arguments you want to make in the prompt. thats about 1/10 of the work of drafting it yourself (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48850565) |
|
Date: April 15th, 2025 6:50 PM Author: ,.,.,.,........,....,,,.. Yes, but that means there’s still a lot of value in judgment to know how to prompt it. It’s not anywhere close to real AI in the sense of being able to figure it out by itself. As long as that is the case there are jobs for lawyers. It will actually likely get more lucrative for successful lawyers. The more value placed on one person to run something the more that person makes $$. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48850574) |
|
Date: April 15th, 2025 10:05 PM Author: Jason Genova I’ve been noticing a significant number of clients emailing me AI outputs regularly now. Never once used chat gtp or whatever it’s called but it has been interesting to see the results that the ordinary prole retard is able to generate (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851296) |
|
Date: April 15th, 2025 11:08 AM Author: ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,, It seems to me that it's still pretty bad at legal research, and none of the legal research platforms built on it are very good. But I agree it's just a matter of time / training data before it's as reliable as an associate. People setting the bar too high--AI doesn't have to be perfect, it just has to be comparable to a replacement level associate for it to be massively disruptive. Associates are often very fucking stupid.
The reasoning models of ChatGPT (o1, to some extent o3) are actually awesome at medical diagnostics and analysis. At least in my experience it sends you down pretty much the same path as a human specialist (as in, suggesting follow up testing, interpreting results with appropriate judgment/caution). It obviously lacks the ability to perceive a patient and observe clinical signs/symptoms at this point and is totally input dependent but still we are heading fast towards a world where standard of care is to consult an AI model. Lawyers will follow quickly. |
|
Date: April 15th, 2025 6:19 PM Author: Aspie Pickup Line Doesn't even have to be as good as an associate, just has to be close enough for partners to be able to correct them in reasonable amounts of time when you consider the cost savings. If it's 90% as good as a junior for 1% of the cost, that's probably ballgame. And you'd have to presume they're improving faster than a human would.
It'll be a big windfall for the partners (who will keep raising bills 6% a year regardless) for a little bit, and then it'll get to the point where the clients can economically onboard the "reviewers" themselves and then any sufficiently large business won't need outside counsel at all. |
|
Date: April 15th, 2025 6:45 PM Author: ,.,.,.,........,....,,,.. I don’t see a near term world where an expert doesn’t need to be involved at some stage of the process. I agree that juniors are fucked. It’s already at the point that they hold little to no value from a research standpoint. They’re just there to make sure emails are ready and things push forward. But AI can already do diligence and research better than them by far (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48850557) |
|
Date: April 15th, 2025 10:19 PM Author: Aspie Pickup Line I don't think anyone at the partner level now is in danger of being replaced in the span of their career. Juniors may be replaced much quicker. Of course, if there are no juniors there will no one with relevant expertise to oversee down the line. But lol at today's millionaires caring about that. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851350) |
|
Date: April 15th, 2025 7:49 PM Author: the walter white of this generation (walt jr.) If it's a *real* 90% *value*, that's correct; if you just mean 90% accuracy or some shit then it's not.
Consider voice transcription, which is still terrible to the point of being completely useless (I try these auto-transcription services regularly, and they aren't worth 10% of the cost of a court reporter). They're 92-95% "accurate", but that level of accuracy is completely worthless and useless.
I've brought this up before, but the fact that court reporters aren't currently anywhere close to being replaced is what's mind-boggling. |
|
Date: April 15th, 2025 10:17 PM Author: Aspie Pickup Line I suspect this "last mile" problem will be the delay in implementation in much of AI. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851333) |
|
Date: April 15th, 2025 6:26 PM Author: metaphysical fallow The issue is it's not getting specifically legal context, it's going off what it knows + google.
Lexis and whatever else are building LLM wrappers around their massive curated databases and will basically google the db and then pass the best results to the LLM with your query.
Once that happens at scale lolers are fkced |
|
Date: April 15th, 2025 10:26 PM Author: hours-hungry BILLPIG Oncologist subspecialist bro said a patient had AI interpret their MRI results the other day and it was perfect (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851385) |
|
Date: April 15th, 2025 10:28 PM Author: metaphysical fallow all professions will get closer to the user. The knowledge layer will crumble. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851398) |
|
Date: April 16th, 2025 12:18 AM Author: ,....,..,.,,,,,.. It uses the o3 reasoning model as the base model for doing research. O4 mini is due to be released soon (openai is claiming this beats all humans at competitive coding benchmarks) so it will probably get better soon too. (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851755) |
|
Date: April 16th, 2025 12:43 AM Author: ,..,,......,....,,,,..,.,... Still plenty of shit law areas where a lawyer is required (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851811) |
|
Date: April 16th, 2025 12:47 AM Author: metaphysical fallow Until law = "Agentic lawyer" corresponding with "Agentic judge" in virtual reality, you're gonna need a lawyer to shepherd lawsuits (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851819) |
|
Date: April 16th, 2025 12:49 AM Author: ,..,,......,....,,,,..,.,... Cr. Transactional "lawyers" deserve what's coming (http://www.autoadmit.com/thread.php?thread_id=5710772&forum_id=2).#48851827) |