5 ms·

Maybe we need a better-scaling legal system that does not take years to resolve simple disputes. I am not convinced AI can replace lawyers, but more people hav

by dozerly 2mo ago

Maybe we need a better-scaling legal system that does not take years to resolve simple disputes. I am not convinced AI can replace lawyers, but more people having access to the law without extremely costly fees does sound good to me.

MichaelZuo 2mo ago
I don’t see how that helps? A lawyer isn’t going to sign on to take liability risk unless they spend quite a lot of time analyzing the AI’s outputs for possible blunders? So it structurally can’t cost significantly less.
- throwaway89864 2mo ago
  Lawyers normally don't have any liability risks, no? And reputational risks are very limited, a lawyer can always point to a judge, to the other lawyer, to circumstances, and away from their own (or the AI's output) blunder.
  - MichaelZuo 2mo ago
    Representing anyone in a court room by definition carries liability risks, because the judge can punish the lawyer unilaterally, but not the AI.
  - watwut 2mo ago
    Layer can be disbarred. And judges do not accept "ai did it therefore I, the layer am not responsible for the false claim" as an excuse from layers.
    - throwaway89864 2mo ago
      Realistically a lawyer never gets disbarred. There is no incentive to disbar a lawyer, and a single client usually wouldn't know if a lawyer is consistant in his/her misconduct, judges can't care less and wouldn't go against their friends/buddies, no?
jay_kyburz 2mo ago
The answer is in the article. There needs to be a financial mechanism to deter nuance complaints. A small penalty for lodging a complaint and losing. This fee should help pay for the courts and reimburse and employer for time spent defending itself. (With the judge / magistrate able to wave the fee on compassionate grounds)
- teddyh 2mo ago
  s/nuance/nuisance/
- simonjgreen 2mo ago
  This is a thing, and in the context of the article too. In UK, if an employment tribunal finds you’ve wasted the tribunals time they can and do order costs paid for both the tribunal and other party. Odd they don’t mention it.
  - BrenBarn 2mo ago
    As with most such things, though, what is needed are escalating penalties. If you're a Joe Schmo who files a complaint that turns out to be invalid, no penalty for you. If you're a rich jerk who files a bunch of such complaints, a few million maybe. If you're a giant company that is using the legal system to push people around, $50 billion.
    - simonjgreen 2mo ago
      Again, this is covered, to an extent. Speaking only for England & Wales, the courts have pretty broad costs powers. Especially if one party is bringing litigation unreasonably. There’s no ceiling to it either, it just has to be defensible. It’s also on the court system to leverage procedure correctly to add safeguards against the kind of behaviour you describe. I understand anecdotally the US system is more open to abuse of this nature, but I don’t know the details. This article though was specific to UK.
- tchalla 2mo ago
  In Germany, the losing party pays 3x the court costs. The only issue is that everyone gets paid - the lawyer, the court etc. The only person that doesn’t get paid for their time investment is the person who is suing.
newyankee 2mo ago
With frontier models, I have found so many self contradicting points in Indian tax law that I would rather welcome a good AI helping to reduce discretionary powers of officers via highly subjective interpretations. This power asymmetry is basically the driver for a lot of developing world corruption.
- perching_aix 2mo ago
  I've been thinking about autoformalizing local laws using agents into TLA+ or something, but it's sufficiently past enough my actual skillset that I'm pretty sure I'd just end up wrestling with slop like a pig in the mud. It's a shame though, I consider law to be just kind of a shitty codebase, with natural language being tortured into cooperating, so it's a really natural fit. I'll probably yield to my temptations eventually and proceed anyways. Lord help me from all the creative but completely detached interpretations I'll land on.
  - arionhardison 2mo ago
    I did this for a few federal agencies, here a few examples https://ice.dhs.dev/program/13732-human-trafficking-investigation https://ice.dhs.dev/program/13732-human-trafficking-investig... https://atf.doj.dev/program/44825-open-gun-store-need-ffl https://atf.doj.dev/program/44825-open-gun-store-need-ffl LMK if you want to know more.
    - perching_aix 2mo ago
      I do, though I'm not entirely sure what am I looking at on those links. Could you start by explaining that? They look like training courses or something. I saw a sequence diagram browsing around, seemed to be specific to a sample scenario?
      - arionhardison 2mo ago
        tl;dr a "program" here is a government program (get an FFL, file a discrimination charge, apply for a benefit), codified so that every step has an actor, typed inputs and outputs, and a citation to the provision that authorizes it. Agents then walk each party through it. And note it points the opposite way from ChatGPT-drafts-your-tribunal-claim in TFA: that dynamic broke because AI made filing free while adjudicating stayed expensive, so the queue explodes. Codifying the procedure attacks the other side; what's actually required, where it actually goes, and whether you have it; before it becomes a hearing in 2030. Fair question, and the "training course" read is not an accident; it's the same shape underneath. A program is an ordered chain of modules, each with a declared actor and typed inputs/outputs. Courses are also that. So it renders with the same components. The sequence diagram you found isn't a sample scenario, it's the deal template's actual step graph; the thing an instance runs on. Three authored files per domain: - an ontology: the domain's vocabulary, its regulatory frameworks with real citations, the O*NET occupations that staff it, the systems of record it touches - intents: what a person actually shows up wanting ("open a gun store, need an FFL"), with typed parameters - deal templates, one per intent: ordered pipeline_steps, each with an actor, inputs, outputs, and a policy_check The page you clicked is generated from the last two deterministically. No model in that path. The part that speaks to your TLA+ instinct: I deliberately don't formalize what the law means. I formalize the procedure, and bind each step to the provision that authorizes it. Formalizing semantics is exactly where you get the creative, detached interpretations you're worried about, because every gap gets filled by the model's guess. Formalizing procedure asks the model to transcribe and cite, which is checkable: - every step input is a ref; param:x, step:3.some_output, system:NICS.event; and it has to resolve. A step: ref must name an earlier step's declared output, so the dataflow is a DAG with referential integrity. - every step's policy_check must name a framework declared in the ontology. A step that no provision authorizes fails validation. So most hallucination becomes a build error instead of a plausible sentence. That's the whole trick. Not a smarter model; a narrower artifact. Concretely, since you're right to expect slop: my first pass at four new agencies came back with 100% of step inputs referencing parameters that didn't exist, and prompts that literally said "Subject?". The validator refused all forty programs. That's the mechanism working; I'd have merged them on a read-through. Intents and flows for ATF, if you want to see the layer under the program page: https://wiki.doj.dev/agent/atf https://wiki.doj.dev/agent/atf Limits, since you'll ask. It decides nothing; no adjudication, and consequential steps are human-gated. It's also not a formal method: the invariants are referential integrity and citation binding, not model checking. The genuinely temporal parts are the deadlines, and those do bite; the NLRB's six-month charge window runs from filing and service, with service being the filer's own duty, so a filing-date-only clock computes the wrong date on a deadline that destroys the claim if you miss it. Re: the sibling comment about discretion; that's the actual pitch. Discretion hides in the gap between the written rule and the practiced procedure. Writing the practiced procedure down, with a citation per step, is what makes the gap visible.
majormajor 2mo ago
Moving from an adversarial each-party-has-lawyers-presenting-as-extreme-a-case-as-possibly-can-be-made-for-their-side system to something more like expert arbitrators and independent/court fact-finders would avoid the 'bury them in paperwork and filings' lawyer-cost explosions. Usually the concerns I've seen are around the lines of "the experts are gonna favor the powerful incumbents who they see a lot" buttttt if our system is already doing that AND costing way more anyway, barring entry entirely to many, is that so much worse? Even assuming we can't try to regulate that?
- eddythompson80 2mo ago
  How is that different from the current system? Wouldn’t you just bury those fact-finders in paperwork? Or are you trying to shift the cost and responsibility of the discovery phase to the court instead? Would the court pass it through to the plaintiffs/defendants like the lawyers do or is it just covered by the state?
  - dghlsakjg 2mo ago
    We have quite a few tribunals, ombudspeople, and other non court dispute resolution services in Canada. Typically, the arbitrator is very well versed in the rules/area they are enforcing, and can see right through a snow job. In my province there is a tenancy tribunal that has to make a ruling before a court gets involved, and the courts rarely disagree with a decision made by the arbitrator in the 1% of cases that do get appealed. Because the rules are so clear, you can shovel piles of paperwork at them, and they will ignore it because they know what pieces of paper they actually need to make their decision. It isn't for complex big money cases, but it means that courts have more time for those cases because they aren't dealing with ticky tacky minor disputes over damage deposits and fender benders. Tenancy arbitration is just one of them, BTW. Just about every industry or agency has an outside ombudsperson that is the first step of dispute resolution before a court will be interested.
    - rob74 2mo ago
      Sounds good, but what's to stop the party with more money to appeal the ombudsperson decision if it's not to their liking and then apply the "we have a whole legal department for this" strategy?
inigyou 2mo ago
AI doesn't give you a lawyer service without fees though. It makes you feel like you're getting a lawyer service, then you show up in court and say what the AI tells you to say and lose ten times your net worth because what the AI told you to say was legally nonsense.
- raincole 2mo ago
  The theoretical system can't be "court." I mean it literally can't. If the cost to file legal cases is actually low then there won't be enough judges to handle them.

by dozerly 2mo ago

MichaelZuo 2mo ago

I don’t see how that helps? A lawyer isn’t going to sign on to take liability risk unless they spend quite a lot of time analyzing the AI’s outputs for possible blunders? So it structurally can’t cost significantly less.

throwaway89864 2mo ago
Lawyers normally don't have any liability risks, no? And reputational risks are very limited, a lawyer can always point to a judge, to the other lawyer, to circumstances, and away from their own (or the AI's output) blunder.
- MichaelZuo 2mo ago
  Representing anyone in a court room by definition carries liability risks, because the judge can punish the lawyer unilaterally, but not the AI.
- watwut 2mo ago
  Layer can be disbarred. And judges do not accept "ai did it therefore I, the layer am not responsible for the false claim" as an excuse from layers.
  - throwaway89864 2mo ago
    Realistically a lawyer never gets disbarred. There is no incentive to disbar a lawyer, and a single client usually wouldn't know if a lawyer is consistant in his/her misconduct, judges can't care less and wouldn't go against their friends/buddies, no?

jay_kyburz 2mo ago

The answer is in the article. There needs to be a financial mechanism to deter nuance complaints. A small penalty for lodging a complaint and losing. This fee should help pay for the courts and reimburse and employer for time spent defending itself. (With the judge / magistrate able to wave the fee on compassionate grounds)

teddyh 2mo ago
s/nuance/nuisance/
simonjgreen 2mo ago
This is a thing, and in the context of the article too. In UK, if an employment tribunal finds you’ve wasted the tribunals time they can and do order costs paid for both the tribunal and other party. Odd they don’t mention it.
- BrenBarn 2mo ago
  As with most such things, though, what is needed are escalating penalties. If you're a Joe Schmo who files a complaint that turns out to be invalid, no penalty for you. If you're a rich jerk who files a bunch of such complaints, a few million maybe. If you're a giant company that is using the legal system to push people around, $50 billion.
  - simonjgreen 2mo ago
    Again, this is covered, to an extent. Speaking only for England & Wales, the courts have pretty broad costs powers. Especially if one party is bringing litigation unreasonably. There’s no ceiling to it either, it just has to be defensible. It’s also on the court system to leverage procedure correctly to add safeguards against the kind of behaviour you describe. I understand anecdotally the US system is more open to abuse of this nature, but I don’t know the details. This article though was specific to UK.
tchalla 2mo ago
In Germany, the losing party pays 3x the court costs. The only issue is that everyone gets paid - the lawyer, the court etc. The only person that doesn’t get paid for their time investment is the person who is suing.

newyankee 2mo ago

With frontier models, I have found so many self contradicting points in Indian tax law that I would rather welcome a good AI helping to reduce discretionary powers of officers via highly subjective interpretations. This power asymmetry is basically the driver for a lot of developing world corruption.

perching_aix 2mo ago
I've been thinking about autoformalizing local laws using agents into TLA+ or something, but it's sufficiently past enough my actual skillset that I'm pretty sure I'd just end up wrestling with slop like a pig in the mud. It's a shame though, I consider law to be just kind of a shitty codebase, with natural language being tortured into cooperating, so it's a really natural fit. I'll probably yield to my temptations eventually and proceed anyways. Lord help me from all the creative but completely detached interpretations I'll land on.
- arionhardison 2mo ago
  I did this for a few federal agencies, here a few examples https://ice.dhs.dev/program/13732-human-trafficking-investigation https://ice.dhs.dev/program/13732-human-trafficking-investig... https://atf.doj.dev/program/44825-open-gun-store-need-ffl https://atf.doj.dev/program/44825-open-gun-store-need-ffl LMK if you want to know more.
  - perching_aix 2mo ago
    I do, though I'm not entirely sure what am I looking at on those links. Could you start by explaining that? They look like training courses or something. I saw a sequence diagram browsing around, seemed to be specific to a sample scenario?
    - arionhardison 2mo ago
      tl;dr a "program" here is a government program (get an FFL, file a discrimination charge, apply for a benefit), codified so that every step has an actor, typed inputs and outputs, and a citation to the provision that authorizes it. Agents then walk each party through it. And note it points the opposite way from ChatGPT-drafts-your-tribunal-claim in TFA: that dynamic broke because AI made filing free while adjudicating stayed expensive, so the queue explodes. Codifying the procedure attacks the other side; what's actually required, where it actually goes, and whether you have it; before it becomes a hearing in 2030. Fair question, and the "training course" read is not an accident; it's the same shape underneath. A program is an ordered chain of modules, each with a declared actor and typed inputs/outputs. Courses are also that. So it renders with the same components. The sequence diagram you found isn't a sample scenario, it's the deal template's actual step graph; the thing an instance runs on. Three authored files per domain: - an ontology: the domain's vocabulary, its regulatory frameworks with real citations, the O*NET occupations that staff it, the systems of record it touches - intents: what a person actually shows up wanting ("open a gun store, need an FFL"), with typed parameters - deal templates, one per intent: ordered pipeline_steps, each with an actor, inputs, outputs, and a policy_check The page you clicked is generated from the last two deterministically. No model in that path. The part that speaks to your TLA+ instinct: I deliberately don't formalize what the law means. I formalize the procedure, and bind each step to the provision that authorizes it. Formalizing semantics is exactly where you get the creative, detached interpretations you're worried about, because every gap gets filled by the model's guess. Formalizing procedure asks the model to transcribe and cite, which is checkable: - every step input is a ref; param:x, step:3.some_output, system:NICS.event; and it has to resolve. A step: ref must name an earlier step's declared output, so the dataflow is a DAG with referential integrity. - every step's policy_check must name a framework declared in the ontology. A step that no provision authorizes fails validation. So most hallucination becomes a build error instead of a plausible sentence. That's the whole trick. Not a smarter model; a narrower artifact. Concretely, since you're right to expect slop: my first pass at four new agencies came back with 100% of step inputs referencing parameters that didn't exist, and prompts that literally said "Subject?". The validator refused all forty programs. That's the mechanism working; I'd have merged them on a read-through. Intents and flows for ATF, if you want to see the layer under the program page: https://wiki.doj.dev/agent/atf https://wiki.doj.dev/agent/atf Limits, since you'll ask. It decides nothing; no adjudication, and consequential steps are human-gated. It's also not a formal method: the invariants are referential integrity and citation binding, not model checking. The genuinely temporal parts are the deadlines, and those do bite; the NLRB's six-month charge window runs from filing and service, with service being the filer's own duty, so a filing-date-only clock computes the wrong date on a deadline that destroys the claim if you miss it. Re: the sibling comment about discretion; that's the actual pitch. Discretion hides in the gap between the written rule and the practiced procedure. Writing the practiced procedure down, with a citation per step, is what makes the gap visible.

majormajor 2mo ago

Moving from an adversarial each-party-has-lawyers-presenting-as-extreme-a-case-as-possibly-can-be-made-for-their-side system to something more like expert arbitrators and independent/court fact-finders would avoid the 'bury them in paperwork and filings' lawyer-cost explosions. Usually the concerns I've seen are around the lines of "the experts are gonna favor the powerful incumbents who they see a lot" buttttt if our system is already doing that AND costing way more anyway, barring entry entirely to many, is that so much worse? Even assuming we can't try to regulate that?

eddythompson80 2mo ago
How is that different from the current system? Wouldn’t you just bury those fact-finders in paperwork? Or are you trying to shift the cost and responsibility of the discovery phase to the court instead? Would the court pass it through to the plaintiffs/defendants like the lawyers do or is it just covered by the state?
- dghlsakjg 2mo ago
  We have quite a few tribunals, ombudspeople, and other non court dispute resolution services in Canada. Typically, the arbitrator is very well versed in the rules/area they are enforcing, and can see right through a snow job. In my province there is a tenancy tribunal that has to make a ruling before a court gets involved, and the courts rarely disagree with a decision made by the arbitrator in the 1% of cases that do get appealed. Because the rules are so clear, you can shovel piles of paperwork at them, and they will ignore it because they know what pieces of paper they actually need to make their decision. It isn't for complex big money cases, but it means that courts have more time for those cases because they aren't dealing with ticky tacky minor disputes over damage deposits and fender benders. Tenancy arbitration is just one of them, BTW. Just about every industry or agency has an outside ombudsperson that is the first step of dispute resolution before a court will be interested.
  - rob74 2mo ago
    Sounds good, but what's to stop the party with more money to appeal the ombudsperson decision if it's not to their liking and then apply the "we have a whole legal department for this" strategy?

inigyou 2mo ago

AI doesn't give you a lawyer service without fees though. It makes you feel like you're getting a lawyer service, then you show up in court and say what the AI tells you to say and lose ten times your net worth because what the AI told you to say was legally nonsense.

raincole 2mo ago
The theoretical system can't be "court." I mean it literally can't. If the cost to file legal cases is actually low then there won't be enough judges to handle them.