🎙 Nelly Pod – Episode Show Notes

Guest Details

Robert Keeling – Partner, Redgrave LLPRobert is a nationally recognized authority on e-discovery, validation, and defensibility. He has led large-scale discovery matters for DOJ and FTC second requests, contributed to the Sedona Conference, and authored influential articles such as Humans Against the Machines and Document Review: You’re Doing It Wrong.

Episode Overview

In this episode of Nelly Pod, hosts Daniel Gold and Brandon Mack sit down with Robert Keeling to explore one of the most pressing issues in modern legal practice: the intersection of AI, ethics, and discovery. Keeling reflects on his earlier arguments about the enduring value of human judgment and examines how generative AI is reshaping validation, privilege logging, and responsiveness review.

The conversation dives into practical frameworks for defensibility, the challenges of authenticating evidence in the age of deepfakes, and the evolving role of specialized discovery counsel. For attorneys, judges, and legal professionals navigating this rapidly changing landscape, Keeling offers both caution and optimism about the future of AI in discovery.

Key Takeaways

• Validation remains essential: Generative AI requires the same statistical validation protocols as TAR, particularly for outward facing uses like responsiveness review.
• Privilege logging is ready for AI: Keeling argues that generative AI has matured enough to safely generate privilege log entries, though edge cases still require human oversight.
• Authenticity challenges loom large: Deepfakes and synthetic media pose unprecedented hurdles for evidence authentication, with family law and employment disputes likely to see early impacts.

Action Items

• Differentiate outward-facing vs. internal-facing uses of AI in discovery, applying rigorous validation only where defensibility is required.
• Document processes proactively and consider sharing protocols with opposing counsel to reduce disputes.
• Begin testing AI-driven privilege logging workflows, while maintaining human “sanity checks” for complex compilations.

Chapters with Timecodes

00:00 – Introduction & Welcome
01:13 – Guest Background: Robert Keeling’s Career and Scholarship
02:44 – Humans Against the Machines: Revisiting Human Judgment in the Gen AI Era
05:27 – Validation & Defensibility: Outward vs. Internal AI Uses
10:12 – Prompt Discoverability and Work Product Protection
13:12 – Privilege Logging: AI’s Promise and Pitfalls
17:34 – Gen AI in Second Requests and Merger Reviews
20:10 – TAR 1.0 vs. Continuous Active Learning vs. Gen AI
21:37 – Discovery on Discovery: Documenting Processes for Courts and Opposing Counsel
23:38 – Evidence Authenticity in the Age of Deepfakes
29:11 – Boutique vs. Big Law: The Future of Specialized Discovery Counsel
32:36 – The Increasing Complexity of Modern Discovery

Compelling Quote

“The prompt is clearly work product, and it’s not even a close question. It reflects the thoughts and mental impressions of counsel about the case. To me, it’s classic opinion work product.” – Robert Keeling