The Case of the AI Brief

Earlier today, Matt Bruenig published Can AI Write A Successful Arbitration Brief? on his Substack newsletter, NLRB Edge. If you'd like to take a moment to read it, you should, and then come back. If you don't have time for that, here's the tl/dr: Matt represented UAW in a case against CSUEU which was submitted to the arbitrator "on the papers," meaning that only documentary evidence and briefs were submitted, and the parties did not call witnesses. Believing (correctly) that he had an easily winnable case, Matt decided to use AI to generate a closing brief based on the documentary evidence and a copy of Elkouri's How Arbitration Works. (As an aside, the newest version is available from the ABA)

Matt edited some extraneous arguments out of the AI brief and submitted it to the arbitrator. He did prevail in the arbitration case, and provided both the award and the brief for reference in his newsletter.

AI use is becoming so routine that it isn't surprising that advocates might be tempted to use it in this way. I have a lot of questions about the wisdom of doing so, and, of course, some opinions. That's what we arbitrators do: have opinions.

• Did he disclose his AI use to his client?

The California legislature (SB574) just passed legislation that the governor signed just a few days ago setting clear standards for lawyers about the use of generative AI. Business and Professions Code 6068.1 now states:

 (a) (1) Nothing in this section shall be construed to abrogate an attorney’s duty to exercise reasonable competence and diligence in the practice of law.

(2) An attorney shall not delegate the practice of law to generative artificial intelligence.

(3) An attorney who uses generative artificial intelligence to assist in the practice of law shall do all of the following:(A) Not enter confidential, personal identifying, and other nonpublic information into a generative artificial intelligence system for which access to confidential, personal identifying, or other nonpublic information the attorney inputs into the system is not restricted to the attorney and persons authorized by the attorney under obligations to protect the confidentiality of the information. (B) Take reasonable steps to do both of the following:(i) Verify the accuracy of generative artificial intelligence outputs, including, but not limited to, the accuracy of all case and statutory citations.(ii) Correct any erroneous or hallucinated output in any material used by the attorney. (C) Disclose the use of generative artificial intelligence to the court for all documents submitted to the court and consider whether to disclose the use of generative artificial intelligence if it is used to create content provided to the public.

While the new law does not explicitly require disclosure to the client, it is implied in the requirement to disclose it the court. Notably, the National Academy of Arbitrators advises arbitrators that they must disclose the use of generative AI pursuant to section 2.B. of the NAA Code of Professional Responsibility. (See Opinion No. 28)

• Did he disclose his AI use to the arbitrator or the opposing party?

Although the new law does not specify as such, I think most arbitrators would expect that the disclosure requirement of 6068.1(a)(3)(C) would extend to an advocate's obligation at arbitration, which would include notice to the opposing counsel.

• If someone submitted an AI brief to me, would knowing beforehand affect my decision?

I think it might affect the trust I have in the advocate to understand their own arguments and the value of any cases they may have cited. I don't rely on other cases to make my decisions, though, and my obligation to the parties is to interpret the contract, not pass judgment on the merits of the advocacy. My main reaction would be disappointment. If an advocate delegates their own intellectual work to AI, that does them and their own client a disservice because of the missed opportunity to deepen their understanding their own collective bargaining agreement, which in turn undermines the long-term relationship between the parties. But my decision would almost certainly be the same.

• Would I, under the same factual circumstances, use AI to write an award?

No. In theory, I understand the temptation to make one's workload less burdensome, but I don't think I can truly understand the CBA without processing the evidence myself. The arbitrator's decision starts with the decision about which facts are relevant to interpreting the contract; delegating even the statement of facts to AI is a delegation of my decision-making authority.

I don't fault Matt for having submitted an AI brief, and I appreciate his candor in sharing his process. I also doubt that Matt was aware of the new CA law, which was literally chaptered less than a week ago. The more AI becomes prevalent in the legal world, the more questions we will have. What are yours?