It seems like the artificial intelligence freight train of the past couple years is finally slowing down, at least in the e-discovery space. For as long as I’ve been blogging in this space (about 18 months now), there has been a pretty steady drumbeat of articles taking one of two diametrically opposed perspectives on AI:

The Robot Lawyers Are Coming for Us All!

or 

AI Is Just Marketing Hype

We’ve all seen these hot takes a few times–and actually, that’s probably part of the problem. Any article on artificial intelligence quickly departs the realm of the practical and real and turns into the author’s musings on technology, human agency, and more. In terms of legal technology, it’s the ultimate Rohrschach test. Whatever we see, it says more about us than about AI. While it might be tough to turn that trend around with just one blog post… I’ll do my best to get the ball rolling.

If we put aside all the hype, and focus instead on getting practical benefits out of the use of AI in e-discovery, it seems like effective use of AI should do at least two things:

  1. It should perform a task that AI is good at. A recent article on AI-generated petitions revealed that AI is probably not the best at identifying things people care about or performing generally creative tasks.
  2. The activity should align with existing e-discovery best practices. Just because an AI is good at something (like playing chess) doesn’t mean that skill will have anything to do with e-discovery.

If you want to understand more about what AI is good at, then I’m going to suggest you download our white paper on Putting AI to the Fullest Possible Use. But if you want to find out a little more about #2 on the list above, read on.

For AI to be truly valuable, they have to be aligned with tasks that hold value across the e-discovery process. Fortunately, there are several valuable use cases where legal technology leverages AI’s strengths to save time and money for in-house legal teams.

Document review

Even with all the advances in review technology, it still remains the most expensive phase of the e-discovery process—and therefore it is a prime target for automation with artificial intelligence. While AI has been used in the review phase of e-discovery for approximately a decade, its current integration into document review has become simpler and more elegant. In the past, document review technology required seed sets, which are samples of the ESI coded by human reviewers to train the AI. Today, deep learning algorithms can operate in the background, observing as human attorneys review documents, learning the criteria that make a document relevant to a particular matter.

Project management

As in the example of Uber, an e-discovery AI can function as a project manager, learning from past actions and results and coordinating them across multiple channels. Similarly, AI can orchestrate the entire e-discovery process. With access to all the relevant data in a given matter, it can make suggestions to improve the speed, cost, and results of a given process.

Early case assessment

By gaining insight before collection, legal teams using AI technology can save downstream costs on hosting, processing, document review, and production, as well as associated staff time. More importantly, though, the legal team gets to the facts of the matter faster, allowing them to make case strategy decisions earlier. In fact, with access to prior case costs and outcomes, the AI can even advise on these decisions!