By Eric D. Morton
Artificial Intelligence is in the news. Congress and the Executive Branch are trying to set guidelines or urge guidelines on creators of A.I. engines (“guardrails” is the term often use). The courts are awash with A.I. related litigation, particularly as to copyright law. There are several lawsuits underway brought by visual artists, writers and musicians over the use of their work in training large language models (“LLM”).
Although the standards and law are not settled as to the creation of LLMs and other issues, users of A.I. or other automated systems are being held accountable for the results. Two cases are illustrative.
Hallucinating A.I. and the Dumb Lawyer
An attorney New York City filed a written brief to oppose a motion to dismiss in a Federal court lawsuit. The brief was entirely researched and written using ChatGPT – a well known A.I. engine. The brief cited several published cases and quotations from those cases supporting the attorney’s arguments. The attorney did no research before filing the brief. The attorney merely took the brief as written and copied onto pleading paper, signed it, and filed it with the court.
The problem was that the cases cited were not real. ChatGPT made them up, including the quotations from them. If A.I. doesn’t know something, it makes it up. These made items are called “hallucinations”. In this case, after the attorney filed his brief, the opposing attorneys researched the cases he cited and could not find them. The opposing attorneys alerted the court. Eventually, the judge in the case sanctioned the attorney, his partner and their firm $5,000.00 each. The judge also granted the motion to dismiss.
Truth can be stranger than fiction. I couldn’t have made this up and I am astonished that an attorney would do this. Without guidelines/guardrails A.I. will give any answer since it is not grounded in source of truth. As someone said: “It’s like your are giving a smart friend a pop-quiz from memory but the friend can’t say ‘I don’t know'”.
Bigoted Algorithm
The Federal Equal Employment Opportunity Commission (EEOC) recently settled a lawsuit against a company for age discrimination. The company was using automated systems to screen applicants for tutoring positions. The systems used by the company automatically rejected female candidates over the age of 55 and male candidates over the age of 60. This was a violation of the Federal Age Discrimination Employment Act. The EEOC file suit on behalf of over 200 applicants who had been rejected. The company and the EEOC settled the case before trial. One notable aspect of this case was that the company may not have known of the prejudice of the systems it was using.
There has been a proliferation of the use of A.I. hiring and other business applications. Furthermore, many companies are contracting with third party vendors to help in recruiting and other business decisions. Those vendors may be using A.I. but the companies hiring them will be responsible for any decisions made by their A.I. tools.
There is also a danger of using A.I. for research, marketing and planning that has prejudices and other flaws that may not be in-line with a company’s goals.
Ignorance is no excuse. The EEOC, the Department of Justice, Federal Trade Commission and various states are all reviewing the legal ramifications of use of A.I. Businesses must understand the nature of the A.I. tools that they and their vendors use. We urge our clients to be aware and not use these tools blindly. Understand the legal standards and your business’ goals
Eric D. Morton is the principal attorney at Clear Sky Law Group, P.C. He can be reached at 760-722-6582, 510-556-0367, and emorton@clearskylaw.com.