I guess the amount billed is newsworthy but I can’t say I’m surprised that it cost millions to administer a very messy probate proceeding with an estate worth 9 figures.
Guest
Anonymous
April 30, 2025 11:43 am
If you’re not using AI for depo prep, motion work, etc. you need to get it together and start implementing it in your practice asap. Borders on malpractice not to use it at this point.
What AI program(s) for law do you recommend? As a colleague, I’ve played around with it but too lazy to go too deep. It’s often easier to just go old school. Serious question.
use whatever is available to you. Try the Westlaw AI for research. But for general writing and edits, try chatGPT, gemini, or Grok. just start playing with it. Give it your most recent email and ask it to improve it for you for starters. Ask it to give you ideas for whatever issue you’re dealing with.
Westlaw AI research is terrible. Every single time it flags a case; the flag is wrong. The questions to AI also return massively incorrect responses. I’m having trouble finding anything good about Westlaw AI
Try limiting your searches on the Westlaw AI to cases from the U.S. Supreme Court and U.S. Courts of Appeals. The results are more accurate when the search excludes district court cases.
I used Grok to review and revise a 40 page Commercial lease agreement (from the perspective of a tenant) and it was invaluable and 95% of my edits were accepted by the LL. Also, what would normally take me 15-20 hours took me about 6. Everybody wins!
This is going to sound like a boomer question but are you using Grok/chatGPT as like an add on to something. Or are you going to the Grok 3 page, pasting in your lease and asking it to review/revise? Sometimes i feel like i’m overlooking some of the more basic tenets.
yes. You can upload the lease document to ChatGPT and ask it to review it and make suggestions for anything that might be problematic. You can ask it to clarify clunky language, etc. You can also upload a complaint and have it summarize it for you have an executive summary before you even start reading it.
121Here.
I copied and pasted ~5 pages at a time (rounded up to the end of the Article), so if something glitched, I had already saved it and wouldn’t lose anything and that made it much more digestible. But, it worked very well for me. I was meticulous and didn’t use everything Grok suggested. But I used most of it for sure.
… but before you rush out and begin using any LLM platform, be sure to read the terms of use to make sure the program can’t harvest your input data, prompts, etc. for its own purposes, as this have pretty obvious ramifications for the confidentiality of documents, compliance with SPOs, etc. There’s even a conceivable path through the work-product doctrine here too.
To be sure. What I wonder, though, is what happens if for some reason GPT gets a subpoena for all information input into its system in a particular case by an adversary. If a user knows their data will be appropriated, then how are their question prompts (and the responses) still covered by the work product doctrine?
Ref: The ABA article on AI.
The takeaway, a judge cannot write a coherent email without AI help? AI research needs to be treated skeptically?
Maybe you are just fooling yourself into believing it saves time, or maybe you are just lazy.
Guest
Anonymous
April 30, 2025 12:19 pm
How have we allowed Ballou to remain on the bench? It’s truly astounding. Aside from her bias and other shenanigans, she lacks basic competence for the job,
Had her on a civil trial a couple of years ago. She did a great job. But I got about 90% of what I wanted, so maybe I’m not objective. She’s obviously on multiple sh*t lists.
I tried a civil case with her recently. She was no worse than any other DC judge I’ve had and better than many. What she didn’t know she went and found out and got the rulings right.
Guest
Anonymous
April 30, 2025 12:57 pm
I do not want Ballou removed from all criminal cases. She is butchering the civil cases that she already has; making her take an all-civil docket will be brutal.
Depends on the case. If the client is an underdog, bad judges are great because they increase the odds and create leverage through unpredictability. That’s why there are no “good” judges and “bad” judges, just situations that are more or less favorable to certain clients.
Same with sports. Increased volatility (e.g., football games played in snow) gives the worse team a better chance.
Guest
Anonymous
April 30, 2025 1:29 pm
My best advice on AI is to really refine your questions. Look at the answers and then redo your question. Of course, check everything. I’ve had great results but have seen it literally make up a case.
1:29 and that’s the point of it right? While i haven’t used AI as much as i would like to, doesn’t it kind of ‘get better’ as you narrow down your points of inquiries. Just like a universal Westlaw question for all jurisdictions will get a massive and uncentered response,, filtering it down with qualifiers, filters word proximity etc will get you where you need to go.
Yes you’re exactly right. It’s applying the research skills we hopefully have to a different tool. Not to be negative at all but I always think that those who complain about it simply aren’t using it correctly. It requires an iterative usage.
Guest
Anonymous
April 30, 2025 1:56 pm
AI is bad for legal research. REALLY BAD. It’s not useful or helpful, but that may change in the future. AI is good for analysis and refining drafting. It’s a tool to help draft documents, it’s not a black box that will do all the work itself.
1:29, 2:19 here. I disagree. Start wide and refine. It can access law journals and cases and really has a deep reach. Give it another shot friend.
Guest
Anonymous
April 30, 2025 3:52 pm
I had the pleasure of litigating with Matt Dushoff very recently. He was so pleasant to work with–always lighthearted and fun to talk to. Sad to hear the news.
I have primarily used Chat to assist in drafting discovery requests. It is wonderful. Of course, I still review for mistakes and to tailor to my case, but it saves a lot of time in the initial drafting phase.
I guess the amount billed is newsworthy but I can’t say I’m surprised that it cost millions to administer a very messy probate proceeding with an estate worth 9 figures.
If you’re not using AI for depo prep, motion work, etc. you need to get it together and start implementing it in your practice asap. Borders on malpractice not to use it at this point.
The boomer attorneys who still use yahoo and hotmail addresses are gonna get steamrolled even more now.
What AI program(s) for law do you recommend? As a colleague, I’ve played around with it but too lazy to go too deep. It’s often easier to just go old school. Serious question.
use whatever is available to you. Try the Westlaw AI for research. But for general writing and edits, try chatGPT, gemini, or Grok. just start playing with it. Give it your most recent email and ask it to improve it for you for starters. Ask it to give you ideas for whatever issue you’re dealing with.
Westlaw AI research is terrible. Every single time it flags a case; the flag is wrong. The questions to AI also return massively incorrect responses. I’m having trouble finding anything good about Westlaw AI
Really?? I’ve used it a handful of times and found it to be amazing (at least way better than digging through cases myself)…
Try limiting your searches on the Westlaw AI to cases from the U.S. Supreme Court and U.S. Courts of Appeals. The results are more accurate when the search excludes district court cases.
I used Grok to review and revise a 40 page Commercial lease agreement (from the perspective of a tenant) and it was invaluable and 95% of my edits were accepted by the LL. Also, what would normally take me 15-20 hours took me about 6. Everybody wins!
I just wrote my third email this week to a client that started with the sentence, “Contrary to what GROK told you…….”
That’s like when I go to the MD and say “Dr google told me…”
This is going to sound like a boomer question but are you using Grok/chatGPT as like an add on to something. Or are you going to the Grok 3 page, pasting in your lease and asking it to review/revise? Sometimes i feel like i’m overlooking some of the more basic tenets.
yes. You can upload the lease document to ChatGPT and ask it to review it and make suggestions for anything that might be problematic. You can ask it to clarify clunky language, etc. You can also upload a complaint and have it summarize it for you have an executive summary before you even start reading it.
121Here.
I copied and pasted ~5 pages at a time (rounded up to the end of the Article), so if something glitched, I had already saved it and wouldn’t lose anything and that made it much more digestible. But, it worked very well for me. I was meticulous and didn’t use everything Grok suggested. But I used most of it for sure.
I know for a fact that the DC law clerks are using it for bench memos.
I know it for a fact because I’ve had rulings that included fake, hallucinated citations.
DC clerks don’t cite check each other’s orders?
Chat is dumber than a box of rocks for legal work imo.
… but before you rush out and begin using any LLM platform, be sure to read the terms of use to make sure the program can’t harvest your input data, prompts, etc. for its own purposes, as this have pretty obvious ramifications for the confidentiality of documents, compliance with SPOs, etc. There’s even a conceivable path through the work-product doctrine here too.
I deleted all personally identifying information, except the locale before copying and pasting.
To be sure. What I wonder, though, is what happens if for some reason GPT gets a subpoena for all information input into its system in a particular case by an adversary. If a user knows their data will be appropriated, then how are their question prompts (and the responses) still covered by the work product doctrine?
ChatGPT Premium is very good
No. Go to the library and hit the decennial digests.
Do not use AI to draft your motions. TF?
Ref: The ABA article on AI.
The takeaway, a judge cannot write a coherent email without AI help? AI research needs to be treated skeptically?
Maybe you are just fooling yourself into believing it saves time, or maybe you are just lazy.
How have we allowed Ballou to remain on the bench? It’s truly astounding. Aside from her bias and other shenanigans, she lacks basic competence for the job,
Had her on a civil trial a couple of years ago. She did a great job. But I got about 90% of what I wanted, so maybe I’m not objective. She’s obviously on multiple sh*t lists.
I tried a civil case with her recently. She was no worse than any other DC judge I’ve had and better than many. What she didn’t know she went and found out and got the rulings right.
I do not want Ballou removed from all criminal cases. She is butchering the civil cases that she already has; making her take an all-civil docket will be brutal.
Depends on the case. If the client is an underdog, bad judges are great because they increase the odds and create leverage through unpredictability. That’s why there are no “good” judges and “bad” judges, just situations that are more or less favorable to certain clients.
This is one of the most intelligent posts I’ve ever seen. You’re probably a good poker player. You WANT bad players at your table.
Same with sports. Increased volatility (e.g., football games played in snow) gives the worse team a better chance.
My best advice on AI is to really refine your questions. Look at the answers and then redo your question. Of course, check everything. I’ve had great results but have seen it literally make up a case.
1:29 and that’s the point of it right? While i haven’t used AI as much as i would like to, doesn’t it kind of ‘get better’ as you narrow down your points of inquiries. Just like a universal Westlaw question for all jurisdictions will get a massive and uncentered response,, filtering it down with qualifiers, filters word proximity etc will get you where you need to go.
Yes you’re exactly right. It’s applying the research skills we hopefully have to a different tool. Not to be negative at all but I always think that those who complain about it simply aren’t using it correctly. It requires an iterative usage.
AI is bad for legal research. REALLY BAD. It’s not useful or helpful, but that may change in the future. AI is good for analysis and refining drafting. It’s a tool to help draft documents, it’s not a black box that will do all the work itself.
1:29, 2:19 here. I disagree. Start wide and refine. It can access law journals and cases and really has a deep reach. Give it another shot friend.
I had the pleasure of litigating with Matt Dushoff very recently. He was so pleasant to work with–always lighthearted and fun to talk to. Sad to hear the news.
A mensch.
I have primarily used Chat to assist in drafting discovery requests. It is wonderful. Of course, I still review for mistakes and to tailor to my case, but it saves a lot of time in the initial drafting phase.