Balkinization: A More Nuanced View of Legal Automation

A More Nuanced View of Legal Automation

Frank Pasquale

A Guardian writer has updated Farhad Manjoo's classic report, "Will a Robot Steal Your Job?" Of course, lawyers are in the crosshairs. As Julius Stone noted in The Legal System and Lawyers’ Reasoning, scholars have addressed the automation of legal processes since at least the 1960s. Al Gore now says that a "new algorithm . . . makes it possible for one first year lawyer to do the same amount of legal research that used to require 500."* But when one actually reads the studies trumpeted by the prophets of disruption, a more nuanced perspective emerges.

Let's start with the experts cited first in the article:

Oxford academics Carl Benedikt Frey and Michael A Osborne have predicted computerisation could make nearly half of jobs redundant within 10 to 20 years. Office work and service roles, they wrote, were particularly at risk. But almost nothing is impervious to automation.

The idea of “computing” a legal obligation may seem strange at the outset, but we already enjoy—-or endure-—it daily. For example, a DVD may only be licensed for play in the US and Europe, and then be “coded” so it can only play in those regions and not others. Were a human playing the DVD for you, he might demand a copy of the DVD’s terms of use and receipt, to see if it was authorized for playing in a given area. Computers need such a term translated into a language they can “understand.” More precisely, the legal terms embedded in the DVD must lead to predictable reactions from the hardware that encounters them. From Lessig to Virilio, the lesson is clear: "architectural regimes become computational, and vice versa."

So certainly, to the extent lawyers are presently doing rather simple tasks, computation can replace them. But Frey & Osborne also identify barriers to successful automation:

1. Perception and manipulation tasks. Robots are still unable to match the depth and breadth of human perception.
2. Creative intelligence tasks. The psychological processes underlying human creativity are difﬁcult to specify.
3. Social intelligence tasks. Human social intelligence is important in a wide range of work tasks, such as those involving negotiation, persuasion and care. (26)

Frey & Osborne only explicitly discuss legal research and document review (for example, identification and isolation among mass document collections) as easily automatable. They concede that “the computerisation of legal research will complement the work of lawyers" (17). They acknowledge that “for the work of lawyers to be fully automated, engineering bottlenecks to creative and social intelligence will need to be overcome.” In the end, they actually categorize “legal” careers as having a “low risk” of "computerization” (37).

The View from AI & Labor Economics

Those familiar with the smarter voices on this topic, like our guest blogger Harry Surden, would not be surprised. There is a world of difference between computation as substitution for attorneys, and computation as complement. The latter increases lawyers' private income and (if properly deployed) contribution to society. That's one reason I helped devise the course Health Data and Advocacy at Seton Hall (co-taught with a statistician and data visualization expert), and why I continue to teach (and research) the law of electronic health records in my seminar Health Information, Privacy, and Innovation, now that I'm at Maryland. As Surden observes, "many of the tasks performed by attorneys do appear to require the type of higher order intellectual skills that are beyond the capability of current techniques." But they can be complemented by an awareness of rapid advances in software, apps, and data analysis.

Surden's expertise in the AI literature is formidable; Simkovic and McIntyre have a similar level of command of labor economics. They come to similar conclusions:

Recent work studying outsourcing and automation has found that work that requires complex thought and cannot easily be broken down into simple rules or algorithms is more difficult to automate or outsource, and this favors highly educated workers such as law degree holders over those with moderate skill and less education (Autor et al. 2003; Autor et al. 2006).

Predictions of structural change in the legal industry date back at least to the invention of the typewriter. Yet lawyers have done remarkably well financially with the introduction and adoption of new technologies and modes of work—computerized and modular legal research through Lexis and Westlaw; word processing; electronic citation software; electronic document storage and filing systems;
automated document comparison; electronic document search; email; photocopying; desktop publishing; standardized legal forms; will-making and tax-preparing software.

That last category may end up providing some unexpected work for lawyers. LegalZoom is in a bit of an identity crisis. The more it tries to go beyond legal form provision, the more it risks unauthorized practice of law suits. Similarly, TurboTax's forums are full of bad advice, much of which strikes me as badly vetted, if reviewed at all. Those who rely on it unwisely may end up dragged into courts or administrative tribunals.

And yet the predictions of doom continue unabated. The real question is: why? For one thing, there's a fair amount of cash in the panic-driven consulting industry. Tell someone that knowledge about automation could help some of their lawyers (or law students) do their job better, and you might get interest, or a shrug. Tell him "THIS COULD BE THE END OF YOUR INDUSTRY AS YOU KNOW IT," and the shrug's less likely.

The Ideology of Labor Devaluation

There also appear to be ideological motives. When restaurant workers demanded a higher wage, hardline owners simply threatened to replace them with robots. Oddly, they didn't seem all that keen on replacing managers with robots, though that could save a great deal of money, too.

The same dynamic applies a fortiori to lawyers, long dismissed as a menace to commerce. Consider, for instance, John O. McGinnis's paean to automation in the libertarian City Journal:

The growing role of machine intelligence will create new competition in the legal profession and reduce the incomes of many lawyers. . . . A decline in the clout of law schools and lawyers could have potentially broader political effects. For the last half-century, many law professors and lawyers have pressed for more government intervention in the economy. This isn’t surprising. Lawyers in the modern regulatory state reap rewards from big government because their expertise is needed to understand and comply with (or exploit) complicated and ever-changing rules. In contrast, the entrepreneurs and innovators driving our computational revolution benefit more from a stable regulatory regime and limited government. As they replace lawyers in influence, they’re likely to shape a politics more friendly to markets and less so to regulation.

For those acquainted with legal realism, “government intervention” is a meaningless phrase. The power now enjoyed by "entrepreneurs and innovators" is epiphenomenal of a long history of law making and enforcement. Government invested in our most important technologies when markets shunned the risk. Some lawyers may "reap rewards from . . . complicated and ever-changing rules" that are, more often than not, lobbied for by the businesses that McGinnis lauds. But without some baseline of rules, standards, and due process intelligible to humans, our economy will only become more divided, more unequal, and less fair.

I will concede one point, though: to the extent conservative courts empower corporations to impose what are essentially feudal terms onto citizens, it will result in a crude simulacrum of law easily parsed by computers. Just as the "creditor is always right" for many credit bureaus, whatever the entity imposing the terms wants (perhaps via unilateral modification) will be the law. Domination makes automation much easier.

Hat Tip: Paul Caron.

*At 9 minutes into the podcast. If anyone can refer me to this algorithm, I'd love to use it in my research.

X-Posted: Concurring Opinions.

Posted 10:20 AM by Frank Pasquale [link]