The EDRM GenAI Survey Results: AI and eDiscovery Trends

Key Takeaways

• Survey of 19 EDRM GenAI Working Group members.
• Text summarization most successful GenAI use case.
• Mixed results for GenAI in document review versus TAR.
• No disclosures to courts; client pressure drives adoption.
• Need validation frameworks and peer‑reviewed research.

Summary

The EDRM GenAI Working Group surveyed 19 senior legal professionals to gauge how generative AI is being applied across the eDiscovery workflow. Respondents, mainly lawyers at large U.S. firms, report strong success with text‑summarization and document‑drafting, while results for full‑scale document review are mixed compared with traditional TAR. The panel highlighted a “first‑draft” model where AI augments, not replaces, human judgment, and noted a lack of formal disclosures to courts. The discussion concluded that the field urgently needs rigorous validation frameworks and peer‑reviewed research.

Pulse Analysis

The EDRM GenAI Survey, released in early 2024, offers a rare glimpse into how a select group of technology‑savvy lawyers are experimenting with generative AI in the eDiscovery arena. Conducted as a 42‑question questionnaire followed by interviews, the study’s 19 respondents represent large U.S. firms and primarily legal‑compliance roles. While the sample is small and self‑selected, it reflects the vanguard of a profession that is increasingly comfortable with AI tools, setting the stage for broader industry trends.

Findings show that generative AI excels at high‑frequency, text‑heavy tasks such as summarization and draft generation, delivering tangible efficiency gains in early case assessment and deposition preparation. However, its performance in document review remains uneven; some firms view AI as superior to technology‑assisted review (TAR), while others find it no more effective than existing processes. The panel stressed that AI currently serves as an “initial draft” engine, requiring human refinement before it can be relied upon in litigation. Unlike the decade‑long resistance to TAR, GenAI enjoys rapid client‑driven adoption, fueled by cost‑containment pressures and the technology’s everyday familiarity.

The survey’s broader implication is a call to action for the legal community. Practitioners expressed concern over the lack of formal disclosures to opposing counsel or courts, fearing work‑product violations. More critically, the industry lacks standardized validation metrics beyond ad‑hoc user satisfaction, underscoring the need for objective, peer‑reviewed research akin to the scientific foundations that eventually legitimized TAR. Establishing rigorous evaluation frameworks will be essential for courts to accept AI‑generated outputs and for firms to mitigate risk while capitalizing on the efficiency promise of generative AI.