The rise of machine learning–driven generative AI has sparked claims that it will radically cut the cost and effort involved in legal work. This optimism rests on the assumption that lawyers can adequately control and mitigate the technology’s risks. Yet recent disciplinary cases in Australia and other jurisdictions—where legal practitioners have submitted false or “hallucinated” AI-generated material to courts—show that this belief is flawed.
This paper argues that a new framework is necessary for evaluating the optimal use of AI in legal practice. The argument rests on two premises: first, that generative AI systems are fundamentally disconnected from factual reality and lack transparency; and second, that lawyers’ core professional obligations—honesty, integrity, and the duty not to mislead the court—are incompatible with unverified AI outputs.
The author proposes an alternative framework called the verification-value paradox. This concept suggests that any efficiency gained from using AI in legal work is offset by the heightened ethical and professional duty to verify AI-produced information manually. As a result, AI’s net practical value to lawyers may be negligible once verification costs are accounted for.
The paper then explores how this paradox affects both professional practice and legal education. It argues that the responsible use of AI in law must be grounded not just in technical competence but in enduring professional values—particularly a commitment to truthfulness and civic responsibility.
Source: Papers.ssrn.com
Leave a Reply