•  
  •  
 

Abstract

As artificial intelligence continues to become an increasingly common tool in the modern legal profession, it is important for the United States to not fall behind in regulating it. Artificial intelligence is being relied upon more and more to support modern trials, but the current Federal Rules of Evidence do not address AI specific concerns, such as a lack of traceability, non-reproducibility, or algorithmic reliability. This regulatory gap leaves attorneys and fact finders uncertain about how such evidence should be evaluated. This Article analyzes the current status of evidentiary rules regarding human and machine-generated expert witnesses and argues that the proposed Federal Rule of Evidence 707 should be amended to better fit the unique demands of artificial intelligence. Specifically, this Article argues that the current draft of the proposed rule does not sufficiently address concerns such as bias and hallucinations. Additionally, there are key distinctions between human and AI experts, such as AI’s inability to be cross examined as a human typically is, that make admissibility determinations difficult. To better address these concerns, this Article recommends amending the proposed rule to include a four-part test requiring the AI-generated evidence to be relevant to the facts at issue, transparency in the use of AI, accurate and detailed record-keeping, and requiring the evidence to be based on sufficient inputs in the use of AI- generated evidence. This Article aims to demonstrate how the recommended amendment would more sufficiently address these concerns as well as promote reliability in the judicial system.

Last Page

286

Download

Share

COinS