When a 5-year-old gets in trouble, they make up every excuse necessary to escape their demise with little remorse for what actually happened. AI is no different, yet it is becoming more present and relied on in daily life. Because of AI’s growing sophistication, lawyers and everyday people are turning to it for their legal troubles. While AI may seem like a promising tool, its vast inaccuracies could leave the legal world more convoluted and inaccessible if handled improperly.
Attorneys are expensive, and the job itself is incredibly time-consuming. Lawyers typically cost between $146 to $554 per hour, making them inaccessible for significant portions of the population. They also have a taxing job and work on average 40 to 66 hours per week.
Because AI is seen as a low-cost and low-time commitment alternative, people with legal needs, as well as attorneys and legal institutions, have turned to it for assistance. For the legal world, AI can bolster productivity to unprecedented levels, and, for individuals who cannot afford attorneys, it can increase accessibility to justice.
Some have already used AI instead of hiring an attorney and even won. A woman named Lynn White consulted ChatGPT to appeal a housing dispute. She fed the chatbot all of the relevant court documents, and at the end of the trial, avoided eviction and paying $73,000. Another woman had an opposing attorney compliment her legal knowledge when she represented herself using AI.
While AI appears promising, damage from its use is already beginning to show. Just like a 5-year-old, when AI has a gap in its knowledge, it can lie and misrepresent the truth. Unlike a 5-year-old, AI delivers its faulty answers with cold confidence, as if they were accurate.
In the legal world, AI will “hallucinate” by improperly citing, misquoting, misrepresenting and even fabricating previous cases. In September, AI was reported to have made mistakes in nearly 60 cases. Most of these errors occurred when individuals represented themselves. Because these people are not legally trained, it is harder for them to catch mistakes made by AI programs, meaning the court or other attorneys have to expend additional resources to identify errors.
At the very least, improper citations can make legal arguments hard to follow. Greater violations, like case law misrepresentation and fabrication, could lead to verdicts based on bad or nonexistent law; AI would essentially “create” law if errors go unnoticed.
Those representing themselves are usually given a simple warning, but many have seen their entire case dismissed. For attorneys, courts have been more strict. One case had six instances of fabricated case law, 25 false quotes and 24 misrepresented case laws. The attorney counsel using AI was forced to pay $24,492 for the opposing counsel’s fees and was warned they would lose their ability to practice law outside of their jurisdiction if the issue occurred again. Many other attorneys have been sanctioned through fines and complaints sent to their state’s bar association.
In addition to damaging individuals, AI misuse threatens the entire legal institution. Research has shown that when AI “hallucinates” research, it can reference that invented research in the future as if it were real. When AI legal hallucinations go undetected, attorneys and other AIs may unknowingly use fabricated precedent. Courts and attorneys may need to perform additional research to pick apart the web of what is real and what is not while already overburdened by caseloads
Attorneys need to press the brakes on AI’s staggering development to assess its risks and proper usage. AI has the opportunity to increase productivity and increase judicial accessibility, but it could also cause irreparable harm to the legal system if adopted prematurely.
