The most prestigious law school admissions discussion board in the world.
|
 |
 
|
AI legal writing is still SPS for anything requiring attention to detail
| jenkem | 11/30/25 |
| chaos is running the show | 11/30/25 |
| ,.,....,..,.,.,,,,..,..,.,..,.,.,.,... | 11/30/25 |
Poast new message in this thread
Date: November 30th, 2025 12:29 AM Author: jenkem You can ask it to churn out a pleadings but it constantly misses critical details, like the fact that two events occurred close in time, or in a particular sequence. It takes me longer to proof and correct for errors than if I had just banged it out myself. Maybe this is useful in probate law or family law, but for shitigation it's a minefield. (http://www.autoadmit.com/thread.php?thread_id=5804303&forum_id=2"#49471419) |
Date: November 30th, 2025 12:30 AM Author: chaos is running the show Notebook is contatly fucking up basic dta inputs now this shit chnges daily it seems you have to be on its ass (http://www.autoadmit.com/thread.php?thread_id=5804303&forum_id=2"#49471421) |
|
Date: November 30th, 2025 12:34 AM Author: ,.,....,..,.,.,,,,..,..,.,..,.,.,.,... It’s interesting because the current models are quite good at getting the details correct for math and coding. Probably because it’s easy to create synthetic data in those domains and then brute force the train the model on shit piles of data. This doesn’t generalize to other domains. They will need to find other tricks to get them to be more precise with legal documents (http://www.autoadmit.com/thread.php?thread_id=5804303&forum_id=2"#49471425) |