Psychopathy in Lawyers

[GoldensRule]

OJ was jury nullification. One black interviewed after the verdict said it wasn't about "justice", it was about " just us".

Just like the 2024 election...

 

[GoldensRule]

AI is not "programmed".

AI is indeed created through "programming" then it is trained.

Your PC was built by humans and so was the Internet. A smart person using both can get real info versus someone who believes hearsay or their own narrative.

Even then, ideologues try to control access to information, such as Woke Google and its doctored search results. Try to search about anything opposing feminist ideology and Google returns pro-feminist propaganda.

AI doesn't have to be perfect to be smarter than nearly all people most of the time, and it's going to get a LOT better.

The "Garbage In, Garbage Out" Rule: If an AI is trained on biased, outdated, or incorrect data, its "intelligence" will reflect those flaws. It cannot "know" a fact is wrong unless it has been given better information to contradict it.

AI does not yet embody intuition or judgement. That is only gained through first hand experience and the making of mistakes. Experimental learning is the next step for AI. If currently one asks an AI to make a judgement call, it will rely not on intelligence but the past judgements of human beings.

 

[adrkwan]

AI could have arrived at the correct conclusion in seconds once being fed all the evidence. Humans can be fooled, misled, and manipulated, especially if lawyers make an effort to find the dumbest and least-educatrd jury pool possible.

O.J. did it, and the jury (criminal trial) was too stupid to figure it out. To this day, some people still think he's not guilty, which is asinine given the facts.

I'm curious if that would actually be the case. Someone should submit the evidence of record to AI and see what it comes out with. We're exploring some of the new products, and I am not aware of any sophisticated early adopters. The only use case I see would be for frequent low stakes disputes that don't require much sophisticated legal work, e.g., cost savings for de minimis disputes like in your Terms of Service.

One example is Arbitrus. https://www.arbitrus.ai/ Some literature on their product and approach. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5115184

It would be a hard if not impossible sell for me to convince management to bind ourselves to an AI determination. I would imagine it would be just as difficult to convince the public to accept that an AI can determine whether or not they lose their freedom.

 

[GoldensRule]

I'm curious if that would actually be the case. Someone should submit the evidence of record to AI and see what it comes out with. We're exploring some of the new products, and I am not aware of any sophisticated early adopters. The only use case I see would be for frequent low stakes disputes that don't require much sophisticated legal work, e.g., cost savings for de minimis disputes like in your Terms of Service.

One example is Arbitrus. https://www.arbitrus.ai/ Some literature on their product and approach. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5115184

It would be a hard if not impossible sell for me to convince management to bind ourselves to an AI determination. I would imagine it would be just as difficult to convince the public to accept that an AI can determine whether or not they lose their freedom.

Considering how so many lawyers believe they can emotionally influence judges and juries and sway them to their side, AI seems like it would be "Too impartial" for the tastes of most litigators and litigants. What's next? AI politicians? an AI SCOTUS? We can only hope, I seriously doubt AI could do a worse job than is currently being done.

On the other hand, we might be closer to Skynet and Judgement Day than James Cameron ever imagined.

 

[adrkwan]

Considering how so many lawyers believe they can emotionally influence judges and juries and sway them to their side, AI seems like it would be "Too impartial" for the tastes of most litigators and litigants. What's next? AI politicians? an AI SCOTUS? We can only hope, I seriously doubt AI could do a worse job than is currently being done.

On the other hand, we might be closer to Skynet and Judgement Day than James Cameron ever imagined.

Yeah, I'm not sure I'd volunteer to be the test case while working out the kinks when the end result could be a loss of my rights, imprisonment, or potential financial ruin. For a parking ticket dispute or other matters under $500, sure.

 

[Jimi Jah]

Just like the 2024 election...

"You win some, you loose some, it's all the same" ~ Jimi Hendrix

 

[Wolkenflieger]

AI is indeed created through "programming" then it is trained.



The "Garbage In, Garbage Out" Rule: If an AI is trained on biased, outdated, or incorrect data, its "intelligence" will reflect those flaws. It cannot "know" a fact is wrong unless it has been given better information to contradict it.

AI does not yet embody intuition or judgement. That is only gained through first hand experience and the making of mistakes. Experimental learning is the next step for AI. If currently one asks an AI to make a judgement call, it will rely not on intelligence but the past judgements of human beings.

Unlike humans which can be diamonds in and garbage out, depending on their big five psych traits, mental horsepower, and their emotional stoicism and intellectual honesty.

 

[xrMike]

Yeah, I'm not sure I'd volunteer to be the test case while working out the kinks when the end result could be a loss of my rights, imprisonment, or potential financial ruin. For a parking ticket dispute or other matters under $500, sure.

You don't have to worry about that, because it would require a Constitutional amendment. And that is NOT going to happen.

We have a right to be judged by a "jury of our peers". Computers and software are not "peers" or equal to humans. Humans created AI, so AI can never be our "peer" or equal to us, or be allowed to judge us.

Trust me, there are more people out there who believe as I do, and would be willing to fight to prevent that from happening. There is no hardware + software stack that will ever judge me, not when it comes to my freedom.

And who's to say that AI cannot also be manipulated behind the scenes, by the Deep State, the Illuminati, NGOs, or whatever dark bugaboo you believe in?

This thread has diverted from the original post. Apologies for my part in that. But to get back on track: Who's to say that some subliminal level of psychopathy CANNOT be integrated into AI by its creators to favor some group over all others?

I'll take a jury of human knuckleheads any day over an LLM I don't understand and can't parse.

 

[adrkwan]

You don't have to worry about that, because it would require a Constitutional amendment. And that is NOT going to happen.

We have a right to be judged by a "jury of our peers". Computers and software are not "peers" or equal to humans. Humans created AI, so AI can never be our "peer" or equal to us, or be allowed to judge us.

Trust me, there are more people out there who believe as I do, and would be willing to fight to prevent that from happening. There is no hardware + software stack that will ever judge me, not when it comes to my freedom.

And who's to say that AI cannot also be manipulated behind the scenes, by the Deep State, the Illuminati, NGOs, or whatever dark bugaboo you believe in?

This thread has diverted from the original post. Apologies for my part in that. But to get back on track: Who's to say that some subliminal level of psychopathy CANNOT be integrated into AI by its creators to favor some group over all others?

I'll take a jury of human knuckleheads any day over an LLM I don't understand and can't parse.

Agree with you 100% on the Constitutional Amendment. I don't feel strongly about many things, but this is one of the rights I do care about.

On the second point, AI itself does already have psychopathic tendancies. It has no remorse or empathy, and it does not feel guilt. Some of the other traits could be purposely or accidentally manifested as some researchers are now finding. https://arxiv.org/abs/2403.15479 It's all fascinating stuff, and what we've just started learning now will go out the window once we move on from weak AI to artificial general intelligence.

 

[GoldensRule]

Yeah, I'm not sure I'd volunteer to be the test case while working out the kinks when the end result could be a loss of my rights, imprisonment, or potential financial ruin. For a parking ticket dispute or other matters under $500, sure.

It would be fascinating to do a few thousand parallel test cases, just to see what the AI results might be.

 

[couragewolf]

Legal knowledge actually a small percentage of what you're playing for in terms of high end legal services. You may be able to get away with rote tasks such as contract drafting or basic legal research. We use quite a few AI legal products and they're not as useful as you'd think, especially when they give different answers with the same prompts.

If you have a recommendation would love to check it out.

I’ve had lawyers structure all kinds of complex business structures for me for sure.

But the role of a lawyer is about to get hyper context specific. If even 10% of the demand for lawyers to write and approve or simple contracts the pathway for junior lawyers to work their way into firms softens, the competition for entry level jobs ramps way up, and overall the employability of lawyers goes Down and therefore the value of their work will be depressed.

The threat of improved specific RAG models making AI a more reliable lawyer for complex advice is undeniably going to continue to pressure the even more complex value ad of the profession. Law is code and LLMs do code well.

Lawyers aren’t alone of course.

 

[Jimi Jah]

Lawyers will continue to do most legal work because the law requires it.

Guess who writes those laws?

 

[couragewolf]

Right, and self-driving cars are already better than humans in aggregate...and it's still early. Who knew someone would be landing rockets on their struts like SpaceX is doing? Who knew EVs would become as popular as they have? Tesla was only formed in 2003. Computers are pretty amazing too, and smart phones. A lot has changed since the founding of the U.S.

An argument from personal incredulity is a familiar refrain, but has been proven wrong so many times throughout history.

Do you think in 10 or 100 years that irrational humans will still be populating juries, the same people who fly into a chimpastic rage if one corrects their repeated misspelling of a three-lettter word? We may not even have juries in 100 years. Hell, we may not have work or jobs in 10-20 years because of AI and robotics. We are about to witness some more major inflection points, and we've already seen computers and the Internet change the world. It's one of the reasons I can work 100% from home.

Why are there Amendments to the Constitution? It's because nobody could consider everything at once back in 1775-6. The forefathers had amazing foresight, but they couldn't have imagined the world as it is today. Few could, especially if they lack insight or imagination.

Forefathers: "What, you let American women vote? But, they don't go to war!".

I respect our founding fathers, but the future is indeed queerer than they supposed or could suppose. That's a paraphrasing of a J.B.S. Haldane quote.

"Any sufficiently advanced technology is indistinguishable from magic." Arthur C. Clarke

To be fair Lockheed basically did this in the 1970s and there was just no fat government contract to incentivize selling it without having to unredact and nullify a bunch of missile contracts

 

[couragewolf]

Lawyers will continue to do most legal work because the law requires it.

Guess who writes those laws?

If lawyers become a rubber stamp for complete work I suspect their ability to charge declines in value

 

[couragewolf]

The last handful of llcs I’ve formed did not involve human beings other than the person in Wyoming or delegate saying “ok”

 

[couragewolf]

I’ve also structured claims against the city I live in I never would’ve bothered with that forced some action I never would’ve considered. Man I wouldn’t want to be a small employer right now suddenly having to navigate highly structured employment law claims either things are gonna get weird. The guys here complaining about being landlords are about to see something else watch

 

[adrkwan]

I’ve had lawyers structure all kinds of complex business structures for me for sure.

But the role of a lawyer is about to get hyper context specific. If even 10% of the demand for lawyers to write and approve or simple contracts the pathway for junior lawyers to work their way into firms softens, the competition for entry level jobs ramps way up, and overall the employability of lawyers goes Down and therefore the value of their work will be depressed.

The threat of improved specific RAG models making AI a more reliable lawyer for complex advice is undeniably going to continue to pressure the even more complex value ad of the profession. Law is code and LLMs do code well.

Lawyers aren’t alone of course.

I don't disagree that AI will replace tasks that are repetitive or rote. Entity formation, tax, or anything forms based is probably going to be automated sooner rather than later. They are not exactly profit centers for firms. From my experience, the bulk of revenue come from high budget legal spend for M&A or some flavor of litigation. For these practices, AI is a nice tool to have but it can't replace even a mid-level attorney.

I don't see the overall employability of the lawyer going down. I see a reduction in supply of new attorneys, like coders now, and an increase in the value of experienced attorneys who can now work more efficiently. The supply of new attorneys has been decreasing long before AI was developed. The damage to your professional reputation for getting caught using AI is also significant. Clients hate it. Judges hate it. It implies laziness whether that implication is warranted or not.

While the law itself may be similar to code, the practice of law isn't.

 

[adrkwan]

Lawyers will continue to do most legal work because the law requires it.

Guess who writes those laws?

We might see some action there in Texas on that sooner rather than later.

 

[couragewolf]

I don't disagree that AI will replace tasks that are repetitive or rote. Entity formation, tax, or anything forms based is probably going to be automated sooner rather than later. They are not exactly profit centers for firms. From my experience, the bulk of revenue come from high budget legal spend for M&A or some flavor of litigation. For these practices, AI is a nice tool to have but it can't replace even a mid-level attorney.

I don't see the overall employability of the lawyer going down. I see a reduction in supply of new attorneys, like coders now, and an increase in the value of experienced attorneys who can now work more efficiently. The supply of new attorneys has been decreasing long before AI was developed. The damage to your professional reputation for getting caught using AI is also significant. Clients hate it. Judges hate it. It implies laziness whether that implication is warranted or not.

While the law itself may be similar to code, the practice of law isn't.

I agree with all of this fwiw, but my position varies in that all the current ai implementations you’re probably seeing are very poorly done and are worse today than you’ll ever see them be again. I suspect that clients of lawyers will become dramatically more difficult to handle as well.

I recently had a very expensive insurance claim challenged and wew lad did I turn that back around on them with a volume of citations their legal dept must have found dizzying

 

[adrkwan]

I agree with all of this fwiw, but my position varies in that all the current ai implementations you’re probably seeing are very poorly done and are worse today than you’ll ever see them be again.

I hope so. I'm in-house now so I'd love to reduce legal spend if I could keep similar outcomes to hiring outside firms. So far, everything has been mediocre at best, and I'm hoping we get some real competition before 2030.