At first I thought this was a fiction story in a hypothetical 2035. The dramatic description of legal technology is completely at odds with my experience. I’m a young, technologically savvy guy who happens to be a lawyer, and I’ve never even used, for example, “AI” discovery. Not for want of economic motivation—I’ve pored through documents manually in contingency cases where having an instant AI answer would have been a god send. But what the article makes it seem like legal tech can do, and what it can actually do, are two things with a vast chasm between.
For example, “AI” discovery requires tons of handholding and careful training/QA. The process is so complicated, it’s really only practical for very large document sets. Ironically, these huge document sets only tend to arise in highly valuable commercial litigation, where clients have strong incentives to demand human second level review anyway, mitigating much of the cost advantage. In practice, the biggest advance in discovery in the last 10-15 years has actually been culling document sets based in search terms (“fake” NEAR “testing results”).
The article also takes a jab at legal outsourcing and makes it seem like AI has had a bigger impact than outsourcing. Not so. The trend of clients demanding that low level work be outsourced to contract attorneys has had ten times the impact on law firm economics compared to any legal technology invented in the last decade.
At bottom, the article is based on a fundamental mistake about the industry:
> What kind of repeat buyer lets the seller push the price up every year for doing the same thing in the same way, where the product stays the same and there is no improvement in the outcome? One answer to that is: a buyer who is faced by a monopoly.
Today’s clients are savvy. They’ll bring in half a dozen firms to pitch any substantial matter, getting dozens of hours of free legal analysis from each just as table stakes. Fees are capped, rates are discounted, and success-based fee arrangements are routine. The legal industry is stiffly competitive, even at the very top. (Thanks to conflicts rules, consolidation in law has been limited. So there are dozens of firms that you would consider just the “top firms”).
On the whole, lawyers keep getting more expensive for the same reason as programmers in Silicon Valley or nannies in Manhattan. The stakes are high compared to the costs, and there is a rush to quality even in light of economic concerns.
I think that AI/Machine learning has made a big impact on the Ediscovery part of the legal industry.
The use of ML commonly known as Predictive Coding does appear to reduce the number of hours required to find relevant documents in a large corpus.
There is a lot of noise about the use of AI in other areas of law, but I suspect (in common with other industries) this is just part of the hype cycle.
That is not to say that the use of AI is not on the rise, but the article likely overstates its use.
Re. Overstating use, far from it. This piece is from 2017, since then adoption of NLP/ML systems has rapidly grown among the leading law firms eg AmLaw 100 and UK 100. Even excluding ediscovery systems the use of AI Tech for review work has hugely increased - sufficient to keep a daily blog going just to keep up with all that is happening. Don’t get complacent. Things among larger firms are starting to change.
Thanks for the detailed comment. Many law firms and in-house counsel are now using AI systems for a variety of tasks, eg m&a doc review. There are review systems that can be focused on providing a risk review for a single document, so it’s not always a scale play. There are about 8 branches of legal AI in use at present in the US and UK, from litigation prediction to some quite advanced expert (rules based) systems.
Clients are pushing back on process work and also looking to law firms and AI providers to give greater insight into their contract stack. As a wise person once said: a business is the sum of its contracts, and NLP\ML systems can provide great insight here, not just cost savings.
Hope that helps.
P.s. the idea that legal costs only increase because clients value the work more is a red herring. The legal market is not truly a market as the buyers at a corporate level are other lawyers who are spending other people’s money (that of the shareholders). But, your defence of the status quo is to be expected by someone who benefits from it being that way.
> P.s. the idea that legal costs only increase because clients value the work more is a red herring. The legal market is not truly a market as the buyers at a corporate level are other lawyers who are spending other people’s money (that of the shareholders).
That's true of all corporate officers. That's why companies put in place incentive structures that align workers' goals with the company's. The idea that in-house counsel are putting the abstract interests of outside counsel ahead of their personal interests in career progression, bonuses, etc., is a very odd one.
> But, your defence of the status quo is to be expected by someone who benefits from it being that way.
This trope ignores the realities of legal billing. There are enormous cost and efficiency pressures in the legal industry. Even for billable-hour matters, clients have a "soft cap" in mind and even the top firms write off 10-20% of billings as a result. Fixed-fee arrangements and hybrid arrangements are also routine. For example, many companies have arrangements where a firm will handle all of the company's litigation in a particular area, or certain portfolios of cases, for a fixed monthly fee. If I'm going to get, say, a million dollars a month to handle a case, regardless of hours, why won't I use some "legal AI" to minimize my investment? If I'm a plaintiff's lawyer, and work entirely on contingency, or doing fixed fee divorces, why wouldn't I leverage every AI tool available?
Your explanation is plausible for the guys billing hours defending Altria from tobacco litigation. But it's laughable as applied to the majority of lawyers.
Also many medium size law firms are using AI systems now. See: https://www.artificiallawyer.com/2018/07/03/tuesday-luminanc...
And,
https://www.artificiallawyer.com/2018/06/19/israels-tadmor-l...
Neither of those two firms are massive, but they both still see the value.
Also AI systems are not as expensive as you may think.
P.s. I’ve worked in the legal sector for 20 years. Well aware of how billing works, and things now are changing esp. re. Any process level work. May I ask which firm you work at?
After reading the article, I came to a different conclusion- that the lawyer writing the article would consider LegalZoom to be AI.
That being said, for many lawyers, the difference is probably indiscernible - too many are stuck in the dark ages.
We fired our corporate lawyer earlier his year because the focus seemed to be on billing hours not solving problems. We found a better and cheaper lawyer who isn’t afraid of piecing together boilerplate to create a rough draft is a short period of time - our previous one had hand written it!!! Insanity.
[content warning: plug for paywalled academic article I wrote]. If the (a bit overheated) claims in this article are true, it probably isn't good for social equality---at least, not in the absence of a sustained effort to build what in effect are social ventures (memo to YC...) around legal AI. Just building machine learning into existing legal workflows will probably reduce the legal costs for the rich and powerful without trickling down to the poor, who often face social in addition to financial barriers to legal services. Article I wrote making this argument at some length: https://utpjournals.press/doi/abs/10.3138/utlj.2017-0047
If the law were automated how would the wealthy get preferential treatment? Entering the legal system rich vs entering the legal system poor are two entirely different things. Maybe you will have to pay for compute cycles or additional law modules or something like that. ...I tend to think control over the law needs to be arrested from human hands if we want a fair system. But people don't want a fair system. They want a system that benefits them. I wonder if AI in the legal system will just be an internal revolution.
I find the article is badly written. Being a lawyer, a coder and interested in Legal Tech, I still find it hard to read or unterstand some sentences/sections. The author also falls short of mentioning any specific workflows that could be automated (which there are many) or any specific programs or companies that are really changing the legal landscape (which there are not many).
From my experience real AI Legal Tech is yet rarely used (at least in the European legal market). And I have yet to see a company I trust that will change that soon.
Thanks. There are over 30 legal AI companies working just in doc review. If you are interested in this area then check out previous pieces that cover a very wide range of use cases. There are well over 50+ companies using NLP/ML in the legal sector. If firms in your country are not then they are perhaps just a bit behind the curve. Come to London and it may change your mind.
I think the author is confusing "AI" with "automation". While the legal industry (among others) has been slow to embrace automation, it has adopted word processing, text searches, and case management software. I don't think true AI is there yet in the legal, or any, industry.
I hope someone will treat the concept in greater depth than I can, but in my mind it seems imperative to universally tax robots and AI for some measure of their income, to balance new distortions placed upon real people's innate capacities to produce. I'm especially interested in a convincing argument of why not to, as none have come past me yet.
Where do we want to be? I'd argue we want full automation and some form of universal basic income (probably implemented through reverse taxation).
Taxing automation over corporate profit makes that reality further away because it lessens incentives to automate. Increasing corporate/estate/etc taxes leaves the incentives to automate intact.
It's like arguing that we should tax renewable energy because we're going to run out of oil eventually anyway.
17 comments
[ 5.7 ms ] story [ 48.6 ms ] threadFor example, “AI” discovery requires tons of handholding and careful training/QA. The process is so complicated, it’s really only practical for very large document sets. Ironically, these huge document sets only tend to arise in highly valuable commercial litigation, where clients have strong incentives to demand human second level review anyway, mitigating much of the cost advantage. In practice, the biggest advance in discovery in the last 10-15 years has actually been culling document sets based in search terms (“fake” NEAR “testing results”).
The article also takes a jab at legal outsourcing and makes it seem like AI has had a bigger impact than outsourcing. Not so. The trend of clients demanding that low level work be outsourced to contract attorneys has had ten times the impact on law firm economics compared to any legal technology invented in the last decade.
At bottom, the article is based on a fundamental mistake about the industry:
> What kind of repeat buyer lets the seller push the price up every year for doing the same thing in the same way, where the product stays the same and there is no improvement in the outcome? One answer to that is: a buyer who is faced by a monopoly.
Today’s clients are savvy. They’ll bring in half a dozen firms to pitch any substantial matter, getting dozens of hours of free legal analysis from each just as table stakes. Fees are capped, rates are discounted, and success-based fee arrangements are routine. The legal industry is stiffly competitive, even at the very top. (Thanks to conflicts rules, consolidation in law has been limited. So there are dozens of firms that you would consider just the “top firms”).
On the whole, lawyers keep getting more expensive for the same reason as programmers in Silicon Valley or nannies in Manhattan. The stakes are high compared to the costs, and there is a rush to quality even in light of economic concerns.
There is a lot of noise about the use of AI in other areas of law, but I suspect (in common with other industries) this is just part of the hype cycle. That is not to say that the use of AI is not on the rise, but the article likely overstates its use.
That's true of all corporate officers. That's why companies put in place incentive structures that align workers' goals with the company's. The idea that in-house counsel are putting the abstract interests of outside counsel ahead of their personal interests in career progression, bonuses, etc., is a very odd one.
> But, your defence of the status quo is to be expected by someone who benefits from it being that way.
This trope ignores the realities of legal billing. There are enormous cost and efficiency pressures in the legal industry. Even for billable-hour matters, clients have a "soft cap" in mind and even the top firms write off 10-20% of billings as a result. Fixed-fee arrangements and hybrid arrangements are also routine. For example, many companies have arrangements where a firm will handle all of the company's litigation in a particular area, or certain portfolios of cases, for a fixed monthly fee. If I'm going to get, say, a million dollars a month to handle a case, regardless of hours, why won't I use some "legal AI" to minimize my investment? If I'm a plaintiff's lawyer, and work entirely on contingency, or doing fixed fee divorces, why wouldn't I leverage every AI tool available?
Your explanation is plausible for the guys billing hours defending Altria from tobacco litigation. But it's laughable as applied to the majority of lawyers.
Also many medium size law firms are using AI systems now. See: https://www.artificiallawyer.com/2018/07/03/tuesday-luminanc... And, https://www.artificiallawyer.com/2018/06/19/israels-tadmor-l... Neither of those two firms are massive, but they both still see the value. Also AI systems are not as expensive as you may think. P.s. I’ve worked in the legal sector for 20 years. Well aware of how billing works, and things now are changing esp. re. Any process level work. May I ask which firm you work at?
That being said, for many lawyers, the difference is probably indiscernible - too many are stuck in the dark ages.
We fired our corporate lawyer earlier his year because the focus seemed to be on billing hours not solving problems. We found a better and cheaper lawyer who isn’t afraid of piecing together boilerplate to create a rough draft is a short period of time - our previous one had hand written it!!! Insanity.
From my experience real AI Legal Tech is yet rarely used (at least in the European legal market). And I have yet to see a company I trust that will change that soon.
Taxing automation over corporate profit makes that reality further away because it lessens incentives to automate. Increasing corporate/estate/etc taxes leaves the incentives to automate intact.
It's like arguing that we should tax renewable energy because we're going to run out of oil eventually anyway.