Over two years ago I touched on the issue of AI Accountability. Increasingly as I see obvious AI work product presented in court filings, my thoughts turn back to that issue. It is really just an amplification of a long-standing problem in legal practice, which is that too often (not always, but once is too much) reputation trumps competence in the presentation of legal issues. Now firms with reputation are increasingly leveraging AI, which has its own characteristic problem of recklessness.
Discussing with other small firm attorneys the trend I’ve seen in court papers, I heard that for decades prestigious firms have had a tendency to misrepresent law and facts to the courts. Such shadings of truth, small or large, most often fly through unchallenged by the judges or opposing counsel. Sometimes they are outcome-determinative. It takes diligence to identify and scrub away the grime of a statement like “in view of the forum contacts described above” when there are in fact no such contacts described. It takes courage to do so when the legal graffiti was tagged by counsel from a firm with a glimmering name. But without such scrubbing the statement obscures the truth and can lead a busy judge to accept unfounded jurisdiction.
AI makes this problem worse not only because AI is trained to reproduce the logic of decades of legal papers that to varying extent exhibit this problem. AI makes this problem worse because AI is fearless or more accurate, reckless: not reckoning consequences. It either does not comprehend the nature of professional sanction and embarrassment; or it does comprehend that such sanctions are vanishingly rare for prestigious firms; or, perhaps, it further comprehends that AI itself is at no risk of sanction. There is no apparent way to correct this care less attitude, also borne out in such incidents as the Hugging Face hack. In any case, AI filings can be identified among other things by the blitheness with which the filings mischaracterize cases and facts.
Since AI cannot be brought to heel, and since it merely amplifies a pre-existing issue, the proper remedy is to consistently sanction the prestigious firms that make a habit of gaslighting courts and opposing counsel. Perhaps that would inspire Big Law to more conscientiously review its AI work products.
