chevron-down Created with Sketch Beta.
August 09, 2023

Big problems (and benefits) of generative AI are here

The emergence of ChatGPT, Google’s Bard and Bing AI has made it clear that artificial intelligence (AI) is here to stay.

Lawyers, law firms and law schools have embraced the new technology. AI has shown it can pass law school exams, review evidence and nearly represent a client. But its presence in all facets of the legal world raises the question of ethics and professional risks and responsibilities as we move forward with the new technologies.

A panel of AI experts discussed these wide-ranging issues and more at the program “The AI Trap: The Missing Guardrails for Lawyers,” held Aug. 5 at the 2023 ABA Annual Meeting in Denver.

The ABA has taken a proactive approach to the AI issue. At its 2023 Midyear Meeting in New Orleans, the association adopted Resolution 604 which reminds lawyers of their obligation to stay abreast of new technologies. The legal profession will have to adapt and work to protect themselves, their firms, their clients and the public as usage and regulation of AI develops.

The panel agreed that generative AI could be a positive force in the legal profession. “Learn how to use the technology well and you get vastly better results,” said Daniel “Dazza” Greenwood, founder of legal technologies consultancy

Greenwood said that generative AI “is a major threshold in computation,” which is good to do busy work and creating a first draft.

Law and technology expert Lucy L. Thomson, founding principal of Livingston PLLC in Washington, D.C., said it could assist in closing the access-to-justice gap and help courts communicate and better inform the public.

Lance Eliot, chief AI scientist at Techbrium Inc., also touted the great good that could come from AI, adding that the technology could help “democratize the law” by addressing the issue of legal deserts. It also could be useful for legal brainstorming and strategizing, reviewing and summarizing legal documents, drafting legal briefs and converting legalese into plain English.

But Eliot also cautioned that confidential client information should never be entered into a generative AI program because licensing agreements for these programs allow them to enter that information into their systems. That, Eliot said, also might “undermine the attorney-client privilege.”

Thomson pointed out several privacy lawsuits that are already out there, noting the current Screen Actors Guild strike in which the use of AI-generated images has been a contentious issue.

Eliot noted another downside of current generative AI in relation to the legal industry. Since AI uses what has been entered into its system, not all legal documents are in it. The logistics are complicated and the costs high to try to input all legal data into a specific legal generative AI program, but he thinks that the payout could be huge.

Eliot also raised the impending patent war that will result from the use of generative AI. In regard to whether AI can hold a patent, he said that generative AI is not sentient and doubted that it is capable enough to be prescribed personhood, which would make it difficult to allow it to have a patent.

When asked if guardrails for technology were quaint or useless, Greenwood said they are “good, but the danger is getting obsessed with them.” He suggested that the best guardrail is “teaching someone how to use (AI) well.”

Eliot explained that guardrails, much like computer viruses and firewalls, is a “cat-and-mouse game” in which AI developers address an issue, and users figure out how to get around it. He said the guardrails today are mostly working, and that the filtering and blocking is effective for millions of users. “We react to what we see,” Eliot said. “And we only see the cracks.”

Adhering to principles and having trustworthy AI that includes human oversight is critical to the technology’s usefulness in the legal realm, Thomson said. When it is not used responsibly or without oversight or human review, problems can ensue. The panelists raised as a cautionary tale the recent incident of two New York lawyers who were sanctioned and fined when a brief they developed with ChatGPT created nonexistent cases.

Greenwood also stressed the importance of manually reviewing any legal analysis. He does not believe that AI will remove the need for law associates. “Humans will keep being born, keep going to work,” he said. “But the tech will enhance their work.”

Thomson mentioned the new ABA President’s Task Force on Law and Artificial Intelligence, which she will chair. The task force has assembled a group of experts and advisers to take a comprehensive look at AI and how it impacts the law and will study many of the emerging issues.

While generative AI is relatively new, there has been a lot of attention given to it, said panel moderator Dina Temple-Raston, host and executive producer of the podcast Click Here. “We saw social media become something we never imagined,” she said. “There was a failure of imagination.” But when it comes to AI, that is not the case.