ChatGpt / Large Language Model Impact on Litigation Forum

[Anonymous User]

Is anyone worried that these AI tools are going to lead to a major reduction in demand for Biglaw legal services? Some of these tools can perform relatively decent legal research on specific and complex topics in minutes and they also are not terrible at applying the law to novel factual scenarios. I am not someone who gets excited about tech at all, but I am really impressed with these models and expect they will continually improve. Current 3L so I don't have much insight on what the practice of law actually entails, but it seems like at a minimum these tools will reduce the need for junior lawyers.

[ProbablyWaitListed]

Have you really found it to be the case they "can perform relatively decent legal research"? I've been playing around with the tools quite a bit because I do think they represent an important development, but have found them completely unusable for legal research because they "hallucinate" (ie completely fabricate) cases, quotes, and other authorities whenever asked for them. And the prose they generate is rather prolix and general.

Not to downplay the development. I used to work as a translator and for years the most efficient way to do that was run it through google translate then check it line by line yourself. Google Translate was good, but GPT4 is basically perfect. As far as I'm concerned, there is no longer any need for human translation of documents. But more generally, as far as legal impact? I think the earthshaking nature is a little overplayed, though yes, I'm sure they will continue to improve.

[jhett]

The current version of ChatGPT will not reasonably replace well researched and well written litigation documents. However, I think it'll only be a matter of time before someone develops an AI chat tool optimized for legal texts. At that point, I expect a fundamental shift in the industry.

[Anonymous User]

Not to be cute, but I generally have a sunny view of the long-term forecast for our industry for as long as lawyers are wildly overrepresented in all branches of government. Anything that truly cuts into legal monopolies will be made illegal, or subject to a lawyer's oversight. It might reduce labor costs for the easiest tasks, but I think large language models are a much bigger threat for non-legal writers and editors than for us.

[charles117]

the great irony of chatgpt is that it only really works if you already know what the answer is. Its too inaccurate otherwise

[calripkenjrjr]

Surprised at the sanguine views already represented here. In my mind the writing's on the wall for juniors in biglaw in ... all practice groups. Within 2-3 years I expect profound change, to the extent that I wouldn't be surprised to see hiring freezes at most major firms by that point?

Btw I would have dismissed what I just wrote as insane like a month ago but have had my mind changed by some of the profound results coming out of GPT-4. Find that "Sparks of AGI" paper and read it if you haven't already. Big changes coming ... I don't think these things are gonna hallucinate fake caselaw forever.

Hope I'm wrong

[ProbablyWaitListed]

Surprised at the sanguine views already represented here. In my mind the writing's on the wall for juniors in biglaw in ... all practice groups. Within 2-3 years I expect profound change, to the extent that I wouldn't be surprised to see hiring freezes at most major firms by that point?

Btw I would have dismissed what I just wrote as insane like a month ago but have had my mind changed by some of the profound results coming out of GPT-4. Find that "Sparks of AGI" paper and read it if you haven't already. Big changes coming ... I don't think these things are gonna hallucinate fake caselaw forever.

Hope I'm wrong

I wouldn't say that my view, at least, is indifferent or completely unworried. Like yeah, this stuff is exciting and could completely rework the entire economy. But it's definitely not there yet, and at least for my present use cases, I've found it to be basically useless outside of the aforementioned translation stuff. I texted my one buddy with legit AI expertise and he said the "hallucination" problem will be solved in a few months. Maybe then it's a different conversation. Just given that right now it doesn't really do anything useful, it's difficult for me to get TOO excited.

EDIT: like for example that sparks of agi paper says "We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span...law, psychology and more, without needing any special prompting." Like right now it can't even brief a case. So I'm not sure where that's coming from. Again, I do think this technology is very exciting and could be revolutionary. But it's tough to get too excited when I try and use it for legal tasks then realize I just have to google it and figure it out myself after GPT4 returns gibberish.

[Pneumonia]

When the electric lightbulb appeared, I wonder if there were many lawyers who said “this might affect other professions, but I haven’t seen anyone use it to read Blackstone.”

[ProbablyWaitListed]

When the electric lightbulb appeared, I wonder if there were many lawyers who said “this might affect other professions, but I haven’t seen anyone use it to read Blackstone.”

I think this is probably in reply to me? Look dude I started paying $20 a month for GPT4 as soon as it came out and have been devotedly playing with it ever since. I permanently keep it open in a tab on my browser window. I also just tried to use it to brief a case and it completely made it up. I've tried to use it for legal and historical research and it just spits platitudes and fabricated quotations at me. I ask it to produce a bluebook citation and it completely screws it up. I understand how earth shaking this could be, but it's hard to say what exactly the repercussions of it will be for the profession when as it currently stands it has essentially no application. The electric lightbulb could have been used to read Blackstone after dark as soon as it was invented. GPT4 right now does not have a legal usage I can see.

[PriyaRai]

Are there any blogs, Twitter/LinkedIn accounts, etc. about practical applications of ChatGPT for legal practice? I’ma litigator and would love to get ahead of the competition (or at least keep pace).

[calripkenjrjr]

Surprised at the sanguine views already represented here. In my mind the writing's on the wall for juniors in biglaw in ... all practice groups. Within 2-3 years I expect profound change, to the extent that I wouldn't be surprised to see hiring freezes at most major firms by that point?

Btw I would have dismissed what I just wrote as insane like a month ago but have had my mind changed by some of the profound results coming out of GPT-4. Find that "Sparks of AGI" paper and read it if you haven't already. Big changes coming ... I don't think these things are gonna hallucinate fake caselaw forever.

Hope I'm wrong

I wouldn't say that my view, at least, is indifferent or completely unworried. Like yeah, this stuff is exciting and could completely rework the entire economy. But it's definitely not there yet, and at least for my present use cases, I've found it to be basically useless outside of the aforementioned translation stuff. I texted my one buddy with legit AI expertise and he said the "hallucination" problem will be solved in a few months. Maybe then it's a different conversation. Just given that right now it doesn't really do anything useful, it's difficult for me to get TOO excited.

EDIT: like for example that sparks of agi paper says "We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span...law, psychology and more, without needing any special prompting." Like right now it can't even brief a case. So I'm not sure where that's coming from. Again, I do think this technology is very exciting and could be revolutionary. But it's tough to get too excited when I try and use it for legal tasks then realize I just have to google it and figure it out myself after GPT4 returns gibberish.

Good points. I might be wrong -- I just know we're in a period of radical uncertainty. I do think though, these things don't have to be 100% on the ball, since associates are not 100% on the ball (the whole business model is built around reviewing juniors' work). Once an AI is 80% good as an average associate, I don't see the business case for partners to keep hiring new associates. Right?

[ProbablyWaitListed]

Surprised at the sanguine views already represented here. In my mind the writing's on the wall for juniors in biglaw in ... all practice groups. Within 2-3 years I expect profound change, to the extent that I wouldn't be surprised to see hiring freezes at most major firms by that point?

Btw I would have dismissed what I just wrote as insane like a month ago but have had my mind changed by some of the profound results coming out of GPT-4. Find that "Sparks of AGI" paper and read it if you haven't already. Big changes coming ... I don't think these things are gonna hallucinate fake caselaw forever.

Hope I'm wrong

I wouldn't say that my view, at least, is indifferent or completely unworried. Like yeah, this stuff is exciting and could completely rework the entire economy. But it's definitely not there yet, and at least for my present use cases, I've found it to be basically useless outside of the aforementioned translation stuff. I texted my one buddy with legit AI expertise and he said the "hallucination" problem will be solved in a few months. Maybe then it's a different conversation. Just given that right now it doesn't really do anything useful, it's difficult for me to get TOO excited.

EDIT: like for example that sparks of agi paper says "We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span...law, psychology and more, without needing any special prompting." Like right now it can't even brief a case. So I'm not sure where that's coming from. Again, I do think this technology is very exciting and could be revolutionary. But it's tough to get too excited when I try and use it for legal tasks then realize I just have to google it and figure it out myself after GPT4 returns gibberish.

Good points. I might be wrong -- I just know we're in a period of radical uncertainty. I do think though, these things don't have to be 100% on the ball, since associates are not 100% on the ball (the whole business model is built around reviewing juniors' work). Once an AI is 80% good as an average associate, I don't see the business case for partners to keep hiring new associates. Right?

Yeah man I agree. The technology is very exciting and has prospects for incredible disruption. I guess I will say two things.

1) if AI is completely able to replace such high skill (ha ha I know, but by any technical definition it is) labor, I think we're gonna be seeing broad-based social upheaval such that a mere contraction in the BigLaw hiring market won't really be at the top of our minds. Sort of like buying US Treasuries. If they go to 0, you're gonna have bigger problems than your investment portfolio taking a hit.

2) Already kinda said this before, but at present the legal application of GPT4 is incredibly limited. I still haven't found anything useful for it, beyond the way it was able to flawlessly translate a French language document I needed. I think some people have suggested it's good at making a text more concise, and I can see some uses for that but still think a human editor is better. Because of its currently limited application, I think speculation about its future impact on the legal industry is putting the cart a little ahead of the horse.

---

For shits and giggles, below is what GPT4 produced when I asked it to make my post more concise. It's cool, but judge for yourself the utility.

"I agree that the technology is promising and disruptive. However, if AI can fully replace high-skill labor, we may face larger societal issues than just the impact on the BigLaw market. Additionally, GPT4's current legal applications are limited, with its main usefulness being translation and potential text summarization. Speculating on its future impact on the legal industry might be premature."