April 2026 — In a recent podcast with the National eDiscovery Leadership Institute (NeLI), Elevate expert Jeremy Pickens, shares insights on how generative AI is reshaping document review while remaining grounded in Technology Assisted Review (TAR).

He explains that many generative AI workflows are, in essence, a return to TAR 1.0, where humans review documents and create prompts that act as static models applied across a dataset. As a result, established TAR principles, such as proper validation and avoiding training on test data, remain as important today as ever.

He further distinguishes TAR 1.0 from Continuous Active Learning (TAR 2.0), emphasising that they are not incremental improvements but fundamentally different approaches. While TAR 1.0 relies on a fixed understanding of a case, CAL continuously adapts, learning from each new document and enabling faster, more dynamic review.

Against this backdrop, Jeremy highlights that success with AI depends less on the algorithm itself and more on how it is implemented. This reinforces the need to evaluate technology based on results rather than features, encouraging legal teams to assess how well a system performs on their data in practice.

Listen to the full episode, here.