Legal technology fully resides in Homer Simpson’s famous vision for Itchy & Scratchy, “whenever Poochie’s not on screen, all the other characters should be asking, ‘where’s Poochie?’” Except “Poochie” is “AI.”
Generative AI looms over every legal tech conversation. If we’re not talking about AI, we are, as Homer suggested, talking about the fact that we’re not talking about AI.
So it came as no surprise that RelativityFest quickly turned to how generative AI tools will enter the eDiscovery space, with the opening keynote announcing the launch of aiR for Review bringing the latest GPT tech to document review.
Chris Brown, Relativity’s Chief Product Officer and one of the few artists capable of pulling off 5 minutes of eDiscovery related stand-up, explained that select customers have been working with aiR and the company plans to have limited general availability at the end of this year.
aiR for Review substantially accelerates various stages of litigation or investigative review, delivering rich case analysis with greater speed and quality. With aiR for Review, users can quickly locate data responsive to a discovery request, material relating to different legal issues and key documents, and view citations and rationale surrounding aiR for Review’s recommendations.
The demonstration focused on the ease of use. Tell the system to review for responsiveness, provide it with a synopsis of the matter, plug in some key terms, give it a sense of what a hot document might be and let it do its work. Important passages get highlighted with the system’s rationales available for the reviewer to evaluate. The user can then add commentary about the result that the system takes to electronic heart and employs in refining the search.
A constant feedback loop.
The reviewers are the brains of the operation and aiR allows the users to see exactly what the AI sees and engage in an iterative process of refining the AI’s approach to the data set. It’s a lot easier and more natural than a cumbersome training regime.
It’s a narrow use case of the technology, though Brown notes that narrowing the purpose also avoids common risks such as misalignment or hallucinations. And it’s also all the space really needs out of this generation of artificial intelligence.
Someday a general artificial intelligence will figure out how to perfectly review documents on its own, right before it launches the world’s nuclear arsenals to cleanse the Earth.
Which is an understandable response to reviewing millions of documents.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter if you’re interested in law, politics, and a healthy dose of college sports news. Joe also serves as a Managing Director at RPN Executive Search.