Artificial Intelligence (AI) is here, love it or hate it. Law offices, like many professions, are currently grappling with the challenges of using AI. Many legal tasks are routine and time consuming, and with a few keystrokes AI can produce drafts and summaries in seconds, as well as reams of legal research.
This can result in time efficiencies for the law firms and potential cost savings for clients, particularly when legal services are billed by the hour. Despite these advantages, AI also presents risks.
Confidentiality
If any details of a legal situation, such as names of parties, are entered into a public AI platform they will be used for the AI’s training data and available to other users. This could result in the loss of attorney client and work product privileges. A secure platform is essential.
Overreliance
Attorneys and other professionals can become dependent on AI, which follows a predictable pattern that is rarely recognized until it is too late:
Stage 1 (Experimentation): Professionals try AI with routine tasks.
Stage 2 (Integration): Professionals become comfortable with algorithmic assistance and begin relying on it for increasingly complex tasks while still retaining their underlying skills.
Stage 3 (Reliance): Professionals struggle to perform tasks without AI assistance.
Stage 4 (Addiction): The skills necessary for independent practice deteriorate—the professional forgets how to function without AI assistance.
AI Biases
An attorney might think that an AI research assistant is objective, but the algorithms don’t work that way. They may favor certain precedents or rely on historical case that that reflect social inequities of their time. Some models systematically disadvantage women and racial minorities. The AI assistant may tell you what it thinks you want to hear, which has led to a phenomenon known as hallucinations.
Hallucinations
If AI doesn’t know the answer to a query, it may predict the most likely answer and present it as a fact. It will invent articles, books, quotes, statistics or fake URLs to support the fabricated “hallucination.”
This is why all AI information needs to be verified. Some attorneys have found this out to their professional embarrassment. There was recently a divorce case before the Nebraska Supreme Court where an attorney submitted a brief containing 57 defective citations out of 63 references. His brief was rejected, and he was temporarily suspended from practicing law pending a full disciplinary hearing.
This Nebraska attorney was not alone in falling for a fantasy. A website titled “AI Hallucinations” has a database compiled by Damien Charlotin that tracks legal decisions in cases where AI produced hallucinated content from around the world.
Clients representing themselves (Pro se) often massively fail to use AI correctly for their legal needs. They are especially vulnerable to AI errors because they don’t have the background or training to understand the nuanced rules and deadlines specific to their legal claim.
Clients will sometimes share AI generated legal research with their attorneys in an effort to help. While it is always good to gather information about matters that are important to you, answers to your specific legal questions are better coming from an attorney than from an AI algorithm.
AI is a powerful tool, but it is not a substitute for competent legal judgment!
