Robot to provide legal advice in traffic case: Should the guild feel threatened?

The DoNotPay app, which bills itself as the "home of the world's first robot lawyer," has announced a stunt that has attracted significant interest: Sometime next month the app will (via smartphone and earbuds) advise an actual defendant in a real American traffic court (not in California although, beyond that, the location is unspecified). This coverage on New Scientist is behind a paywall; Lauren Leffer's Gizmodo article is not.

Ms. Leffer's article notes that, "DoNotPay is also reticent to disclose case specifics because what they’re doing is likely in violation of courtroom laws and protocol."

Not just "likely," actually; it's a dead-cinch certainty in Illinois and probably in most places. Apparently, though, DoNotPay has found a jurisdiction which allows hearing impaired persons to use phones with Apple AirPods to amplify the proceedings. If Mr. Roboto (that's a Styx reference, not a sexist lapse) ever wanted to be a real lawyer, he might have a hard time with Character and Fitness should these shenanigans come to light. Ms. Leffer's article quotes DoNotPay's CEO, Joshua Browder, as saying the court will "definitely not" be aware of these arrangements -- so the company is setting out to deceive the court.

And, of course, actually whispering what to say in the client's ear would get any human lawyer sanctioned on the spot.

Leffer's article says DoNotPay is also planning a second stunt involving a speeding defendant facing a Zoom trial. "In that instance," she writes, "DoNotPay is weighing the use of a teleprompter vs. a synthetic voice — the latter strategy Browder described as 'highly illegal.'" Actually, neither strategy would be open to a human lawyer. At least more than once. If said lawyer were caught.

So these tests (or "experiments" as Browder calls them) would appear to confer a huge, as well as unfair, advantage for the AI over a similarly-situated human attorney.

But I wonder....

It seems safe to assume that the human test subjects that DoNotPay has recruited for these exercises have been carefully screened; like news anchors or sports broadcasters they would have to relay the AI-provided information seamlessly, without any perceptible delay as new instructions are received. Moreover, these first subjects would be strongly inclined to follow the direction provided by their robot lawyers. (And incentivized to follow that advice, too: DoNotPay promises to pay any fines that may result from adverse determinations.)

Having a direct, albeit surreptitious, audio channel into a typical client's cranium might prove a more mixed blessing for man or machine alike.

But let's put that aside for a moment.

A post by Anugraha Sundaravelu about these "experiments" on Metro.co.uk, claims that Browder's "ultimate goal is to have his app replace lawyers altogether in order to save defendants money" (the DoNotPay website is careful to note, on the other hand, that "DoNotPay is not a law firm and is not licensed to practice law").

But to determine whether non-lawyer robots could really replace living, breathing lawyers, said robots should be subject to the same strictures as the people they would supplant. So -- for a fair test -- no whispering in the defendant's ears. The robot and the defendant can chat outside the courtroom, as people do with their human lawyers, but not inside the courtroom. And certainly not after every question asked.

Real, unscreened clients routinely ignore their human counsel's advice and instructions -- at least until things go south -- at which point the lawyer is usually blamed for failing to provide what he or she has desperately tried to provide right along. Would robots really fare better?

I'd bet against it. And a disembodied AI can not even hunch over and cover its face in frustration as the client torpedoes his or her own case.
For the foreseeable future, at least, lawyers should have no fear of AI-replacements.