Judges in England and Wales Given Cautious Approval To Use AI in Writing Legal Opinions

Press2ToContinue writes: England's 1,000-year-old legal system -- still steeped in traditions that include wearing wigs and robes -- has taken a cautious step into the future by giving judges permission to use artificial intelligence to help produce rulings . The Courts and Tribunals Judiciary last month said AI could help write opinions but stressed it shouldn't be used for research or legal analyses because the technology can fabricate information and provide misleading, inaccurate and biased information.

"Judges do not need to shun the careful use of AI," said Master of the Rolls Geoffrey Vos, the second-highest ranking judge in England and Wales. "But they must ensure that they protect confidence and take full personal responsibility for everything they produce." At a time when scholars and legal experts are pondering a future when AI could replace lawyers, help select jurors or even decide cases, the approach spelled out Dec. 11 by the judiciary is restrained. But for a profession slow to embrace technological change, it's a proactive step as government and industry -- and society in general -- react to a rapidly advancing technology alternately portrayed as a panacea and a menace.

So Much for Case Load

[packrat0x]

AI written text will waste the parties' time and the court's time. It will make cases move at a glacial pace.

Not all technology is relevant to the court.

Re:

[zenlessyank]

But how much time will be saved by AI not having to dress up in those silly outfits?

Re:

[EvilSS]

Oh no, the AI still has to wear a powdered wig to court.

Re:So Much for Case Load

[DarkOx]

Using AI to draft up opinions isn't the worst idea.

The *gpt seems to do a pretty good job at taking a list and turning into prose; and ones with few grammatical or typographical deficiencies at that.

As long as people read it over for content and make sure nothing was added or any critical elements were dropped, there should be no real issues with this. i can probably do a passable job doing things like changing simple citations ie "Bush v. Gore" into the full citation. Well enough that an legal scholar reading the output over can confirm "yeah that is accurate" without having to fully look it up, that is if it dates in 1972 or something they'd know immediately that AI got it wrong.

It could be good tool, for drafting opinions and briefs if it is used as a drafting tool, and not "write me an opinion where blank .. is found to be equivalence to blank.." low effort prompts are going to lead erroneous results for sure.

Now a better question is do you need or even really want generative - general purpose AI to do this job? More traditional formatting and transformation and lookup techniques, rolled into a tool specically targeted at legal briefs might very well generate just as nice a produce with a lot less risk, a lot fast, and resource consumption. However "just call the chatGPT api" is easy for developers.

Re:

[phantomfive]

The *gpt seems to do a pretty good job at taking a list and turning into prose; and ones with few grammatical or typographical deficiencies at that.

Not brief prose.

Re:

[serviscope_minor]

That I think illustrates the problem with AI and a deeper problem.

If you can write a brief, but complete draft as a list, it seems pretty pointless to expand it to be longer just so a human can take longer reading it to form a mental list with all the points.

Re:

[classiclantern]

This is only the first step. Look at the goal. A legal system where AI could replace lawyers, help select jurors or even decide cases. What's next? AI powered governments? Don't say I didn't warn you.

Tramslation

[Rosco P. Coltrane]

The justice system is so backed up that we'll use anything to expedite judgements.

Re:

[Fly Swatter]

It is a problem entirely of their own making.

wat

[Pseudonymous Powers]

[carefully and thoroughly removes every trace of bathwater from the baby]

[hucks baby out window]

Re:

[omnichad]

Perfect metaphor

Worst possible use...

[Junta]

In this context, all generative AI does is to make the legal opinion overly complicated. This is something the legal community should be trying to alleviate, not make even worse by offloading it to generative AI. Whatever that prompt you would use to describe the ruling? Just describe the ruling directly that way.

At least with searching/analyzing case law, you could require attributions and say those *must* be followed, but the search itself may alleviate workload and perhaps be more effective than tradi

Re:

[account_deleted]

Comment removed based on user account deletion

not bad

[noshellswill]

JapeChat   machine ( volt/amp/watts .. remember  Bosco ) making legal decisions  that could drive a human into the mud ? Not just a bad  skihtzo byteboi idea ... but horrid. ROMAN JUSTICE baby to all who support that bastardization of justice. 

Will the AI lose evidence ?

[Mirnotoriety]

Will the AI detect when the prosecution loses evidence that could exonerate the defendant?

So who's opinion is it?

[Fly Swatter]

If this was high level dictation then sure it would be the Judge's opinion, but if anything is pulled from external sources by the AI then it is NOT the opinion of the Judge; call it what it is - a citation rubber stamped by the Judge.

Although if it could replace a judge along with all the lawyers then THAT would be a win.

Actually Sounds Sensible

[AlanObject]

We are evolving here, and it seems that the judges in there have at least some idea of what they are dealing with.

All the hype around AI and LLMs have done a lot of damage. A superficial look at it -- which is what most media does most of the time -- thinks they are looking at Lt. Cmdr Data when in fact what they are looking at is closer to a Speak-and-Spell (tm)

Maybe they started to realize the obvious: It Is A Tool. Not an Oracle and not a sci-fi calculator to the Answer To the Universe (which works out to 42.)

Use the tool smart/skilled you get smart/skilled results. Use it like a clueless idiot and get clueless idiot results. Like any tool!

Is that so hard?

Re:

[Junta]

Think it's more sensible to use it in the context of enhanced search (which they said is forbidden) and ban it for creating new ruling material (where it can *only* serve to make things more wordy and thus harder to understand, compared to the prompts they would provide that may be easier to read and understand than the formal BS legalese).

I guess it will have to be a Lord AI then

[nospam007]

Like most of the top judges and police brass.

the next brief

[sdinfoserv]

Be careful what you allow - The next legal brief might argue the cost savings of eliminating judges in favor of legal AI bots. just like in Idiocracy

Re:

[bn-7bc]

Nah that would not be LeagalBot but HRbot and AccontantBot the judges would just contribute to the right campaigns and get ,legislation in place saying somthin like " no judgements or opinions above the severity of small clames court shall be given/passed by AI"

Already have suggested sentencing macros

[Canberra1]

Yep, the sentence is often auto suggested to the Judge. Factors include if all the jails are full, or if child support is maxxed out. So in Australia a criminal with 9 children born to six fathers, someone has to care for the feral kids and this is very expensive. So the record is being given bail six times on the SAME day. There is one judge who relies on written submissions only. At court they are asked if anything has changed. If one side is unprepared and tries to 'oral' it out, he/she is told off, and