As generative AI continues to reshape eDiscovery, one question consistently rises to the top: Can legal teams truly trust AI-driven review results? In the latest episode of 5 in 5, CDS’ VP, eDiscovery Products & Solutions Michael Milicevic sits down with Relativity’s Senior Manager, Applied Science Nathan Reff, Manager of Applied Science at Relativity, to tackle that question head-on and explore how aiR for Review ensures its results are accurate, defensible, and consistent. If you’re curious about what “verified AI” really looks like in practice, this is a must-watch conversation.
In just five minutes, Michael and Nathan break down how Relativity designed aiR for Review with defensibility at its core. The episode gives viewers a peek into how the tool confirms the citations it surfaces, the validation methods behind its performance, and why transparency is central to adoption in the legal space. Rather than treating AI as a black box, aiR for Review is built to show its work, backing up decisions with grounded evidence legal teams can confidently stand behind.
Nathan also touches on the metrics that matter most in AI evaluation, and why repeatability and reduced variability are critical for defensible workflows. He and Michael discuss how Relativity approaches verification and consistency, and how the right instructions, review protocols, and sampling strategies help aiR achieve reliable outcomes across matters and data types.
One of the most interesting parts of the discussion is the balance between human expertise and AI-powered acceleration. aiR for Review isn’t designed to “learn as it goes” in the same way TAR does, and the episode briefly explains why that’s actually a strength for teams who need defensibility, control, and predictability in their process.
If you’ve been wondering how aiR for Review delivers results you can trust—and what really goes into validating those results—this episode pulls back the curtain just enough to get you thinking, while leaving plenty to explore in the full conversation.
👉 Watch the full 5-minute episode to hear Michael and Nathan break it all down.
