I just learned of new federal rules of procedure that made my stomach turn. > Like all evidence, the proponent must establish authenticity and admissibility. T

I just learned of new federal rules of procedure that made my stomach turn. > Like all evidence, the proponent must establish authenticity and admissibility. The issue of authenticity has been addressed in federal court with the addition of Fed. R. Evid. 902(13) and (14). Louisiana has not yet adopted similar amendments to the Louisiana Rules of Evidence. > This rule eliminates the necessity of a foundational witness at trial. The person who makes the certification must meet Rule 602 (personal knowledge) and Rule 702 (scientific or specialized knowledge) and Rule 901(b)(9) requiring explanation of how the process or system that generated the electronic record produces reliable and accurate results. > With a human expert, if you want to challenge the opinion, you file a motion in limine which allows you to put on evidence prior to trial to contest the expert's qualifications to give the opinion or the expert's methodology. > The proponent of the AI opinion will have to establish that the Al algorithm produces accurate results. If the proponent cannot do that, then the evidence is unreliable. Unreliable evidence is not relevant. [...] > At the end of the day, if the court finds the Al opinion authentic, then it will be admitted in evidence and the opponent will not be able to cross examine the opinion at trial. Gosh, the only thing I can think of that might save us is challenging the features that become inputs into the models - however, with GPT-4's example, as those increase to inhumane sizes, I wonder if even that will remain an effective strategy. Source: Louisiana Bar Association Journal