It started harmlessly enough, with kids using artificial intelligence to cheat on their writing assignments, but the technology has become a palpable threat to society as lawyers and others in the justice system have conducted novel experiments with it and even clearly misused it.
In the five years since OpenAI unleashed Chat GPT-3 on the public, people have found creative and sometimes unwise uses for the technology, including attorneys who harnessed it to write briefs with fake citations.
Recognizing the risk, the Georgia Supreme Court undertook a 10-month review in August and released new recommendations on Thursday. The state’s high court proposes a three-year process to adapt to AI.
It will start with establishing leadership and governance and conclude with new policies and processes for all the courts in Georgia’s judicial system. There will be community engagement, process reviews, education and training, and the establishment of business and technology architectures along the way.
The committee behind the new report, “Artificial Intelligence and Georgia’s Courts,” was led by Justice Andrew A. Pinson. It incorporates observations by the State Bar of Georgia’s Board of Governors, who produced their own report on the risks of AI in early June.
The bar’s report said revisions to a rule of conduct for lawyers was “particularly critical” because it was about their competence and proficiency with technology.
“It is the committee’s assessment GenAI tools will in short order become ubiquitous,” the bar report’s authors wrote.
Pinson’s committee cited numerous examples of AI uses that occurred just during the 10 months of their review process, such as the Indiana Supreme Court’s introduction of AI for voice-to-text transcriptions, the Arizona Supreme Court’s use of AI avatars to deliver news about rulings by their justices, and a family’s use of AI to create a victim impact statement by their dead relative during the sentencing phase of the trial over his road rage death.
“A key challenge the committee faced during its work is the rapidly evolving nature of a technology new to courts and organizations across the country,” Pinson’s committee concluded.
The panel noted acceptable uses for AI such as for research and scheduling, unacceptable uses such as for jury selection and “black box” sentencing algorithms, and potential uses that need more study and testing such as language translation and sentencing and risk assessments.
