An apparently novel decision in Canada setting aside an arbitral award for overreliance on artificial intelligence raises several important questions, including: How much is too much reliance on AI, and in the absence of glaring issues like hallucinations, how would you even be able to tell?

The set-aside decision at issue came from a trial court in Quebec, Canada, in a dispute involving a healthcare facility. Judge Martin F. Sheehan concluded in his April decision that the award should be set aside due to the fact that 100% of the authorities cited in the arbitrator's analysis had been hallucinated—fabricated—by an AI tool.

The fact that the hallucinations were so blatant made this case more unusual, and probably made the award easier to set aside.

"What it came down to was that the arbitrator really wasn't diligent, and wasn't following current practices as far as using AI and checking citations, so to a certain extent, I think the employment of AI in that case is a little behind the curve," said Duane Morris LLP partner Jeffrey L. Hamera. "So it's kind of an anomaly, but it's still a fundamental problem."

"AI is being used, and is going to be used in all areas of legal work, and it's really a question of, how do you use it reasonably?" he continued. "Arbitration is a little tricky because … typically arbitration awards are given a lot of deference by the courts. They're hard to stop. Once the award is out there, it is usually enforced."

To read the full text of this article, please visit Law360 (subscription required).