Computational Law Applications and the Unauthorized Practice of Law

In the “Siri: What’s Next” post, Siri weighed-in on my used-car purchase decision by providing advice on the likelihood I would find a better deal on a warranty. At the time I purposefully left unattended the question of whether this constituted legal advice and, if so, whether it amounted to the unauthorized practice of law. I’ll begin touching on this intriguing question now.

Let’s begin with addressing why unauthorized practice of law (UPL) even matters here. The bottom line concern here is that if the delivery of legal services by AI is interpreted as UPL, it will likely spell the death knell for any significant development for computational law AI applications (CLAI). Developers who nevertheless dare challenge the UPL rules will need to be well prepared to face the prospect of being dragged into turbulent and expensive legal fist-fights. Will it be worth the bother? Perhaps. But rather than adopt a wait-and-see attitude, a more compelling challenge is to figure out what measures need to be taken, what concerns need to be ameliorated so that we reduce the risk to this valuable area of development.

It’s important to note that I am not suggesting that UPL rules are redundant. I think they serve an important purpose, which can be distilled to the following principles: protect (a) the public and (b) the integrity of the legal system from the incompetence of non-lawyers. (Collectively, the “Rule.”) Statutes, bar rules and common law, maintain the UPL system and the courts usually have the final say, though deferring to Congress.

As unfortunate as it may be, it is of scant surprise that in such a maintenance-heavy system, the UPL landscape is perforated with inconsistent and/or ambiguous definitions. Over all, this malaise dilutes its efficacy, renders compliance unnecessarily challenging and (likely) stifles precisely the CLAI innovation I am concerned about here. If we can achieve a relative degree of clarity in terms of how the UPL applies to CLAI, i.e., defining the “un-crossable thresholds” (UT) we will have made a few steps toward better securing a future in which significant development will be found.

Initially, at least, the UT can be defined in terms of what it is not, where it is not deemed to exist. In that sense, there should be no UPL difficulty/tension in the wide array of areas lawyers could provide legal opinions on, but for a variety of reasons do not. Siri’s opinion on the used-car warranty is a useful example, so let’s continue with it.

The mere act of collating data about used-car warranties is not in and of itself the practice of law (cf. the LegalZoom experience). However, delivering an opinion that a particular warranty from amongst that collection is “better” arguably is. But the next question we must ask, which is a harm-centric one, is: So what?

It is difficult to persuasively argue that it would neither be desirable nor useful to understand what type of warranties are out there. Even so, prospective used-car buyers virtually never hire a lawyer to conduct a warranty compare-and-contrast exercise. It’s too expensive. It’s too time-consuming. CLAI, in contrast, is perfectly suited to deliver this type of service. It can do so very efficiently (i.e., accurately and quickly), especially when it can run a comparison against a large warranty database, one that is well accepted as being representative of other warranties in, for example, the used-car market.

Aside from steering away from areas lawyers do not serve, another (but not exclusive) approach is to set the following CLAI design and operational principle: So long as the CLAI’s operation does not compromise the Rule, it should, under most circumstances, be exempt from UPL scrutiny. Whether or not an exemption is granted could be, for instance, a function similar to the vetting performed by Apple as it filters through which apps will be accepted into the App Store. (Please regard this as a rough sketch, there is clearly more to elaborate on this vetting point alone.) Whether or not the CLAI deviated from this principle could still be the subject of judicial review, however the CLAI designer’s liability would/should be tempered by the fact the CLAI was vetted by a third party.

To further provide comfort, at least in the early phases of molding these CLAI-centric UPL principles, is to require that the CLAI be jurisdiction-sensitive. Given that virtually all smartphones can “tell” where they are, in jurisdictions that insist on forbidding certain CLAIs the app would be temporarily disabled, until (if and when) the app user would travel to a jurisdiction in which it was permissible. While this is a potential inconvenience for users, it can serve as an important interest-balancing mechanism. It is also important to keep in mind that to the extent that such prohibitions exist, they will probably be short-lived as it is reasonable to expect that in a matter of a few years, the UPL attitudes toward CLAI applications will be Federalized.

Tags: