AI, Accountability, and the Future of Discovery

Guest NotesJames Sherer is based in New York, where he co-leads the Emerging Technology Team within the Digital Assets and Data Management (DADM) Group at BakerHostetler. With a unique background that includes a J.D., an MBA, and serving as an in-house litigator for a Fortune 500 company, he provides commercially focused advice on the intersection of law, data, and advanced technology. James A. Sherer - BakerHostetlerKey TakeawaysEthical duties still apply, even as AI changes fast. Lawyers remain responsible for anything produced under their license; AI does not displace duties of competence, diligence, candor, or confidentiality.Privilege and discoverability depend heavily on how the tool is used. Emerging case law shows that courts are not treating all AI-assisted work product the same, especially where confidential litigation material is entered into outside tools.Agentic AI will almost certainly become part of discovery. Prompts, chat histories, logs, model choices, impact assessments, and related governance records may all become relevant sources of evidence.Explainability and governance matter as much as output. If organizations want the benefits of AI systems acting on their behalf, they must have documentation, evaluation, maintenance, and defensibility. Hallucinations and misuse are a supervision and accountability problem as much as a technology problem.Vendor dependency is a major hidden risk. Organizations may deploy AI tools without having contractual rights to access the technical information they later need for audits, litigation, or regulatory scrutiny.Knowledge loss is becoming a governance problem. As systems are implemented and the people who built or managed them leave, organizations may lose the human context needed to explain why a system exists, how it works, or what data it touches.The profession is under real pressure to adopt AI quickly. Market demands to “do more with less” are pushing firms and companies to use AI at scale, often before governance, training, and controls are mature. Technical competence is a continuing obligation, not a one-time certification. Lawyers should approach AI with the same rigor used in litigation, discovery, and expert-driven matters.For future lawyers, the message is ownership plus curiosity. Sherer urges law students and young lawyers to remember that their license is their own, and that long-term success will depend on disciplined learning, visible value creation, and adaptability.