How Active Learning Helps Attorneys Validate Using AI for eDiscovery

By Sarai Schubert

In a recent interview with eDiscovery Today, I mentioned that AI tools have been slow to gain traction in the legal industry because software providers—including IPRO—weren’t making them easy enough to understand or adopt. Let me explain.

Consumer applications that use AI, like Netflix or Amazon, have become mainstream because they’re simple for people to grasp how they use machine learning in the background to offer recommendations based on past use.

However, we instead made AI-powered solutions for the legal industry too confusing to comprehend, where attorneys weren’t able to defend, describe, or collaborate with opposing parties around the need to use the technology in their processes and es a result, they ended up shying away from it.

As technology providers in the Legal industry, we should seek to augment a legal team’s current workflows, not try to change them completely.   

Over the past few years as early adopters have begun to defend using AI in their process with the courts, judges are now accepting the use of AI when it affects proportionality. However, you still have to get opposing counsel onboard and validate each other’s processes to ensure you didn’t miss anything.

What happens if an attorney already negotiated a list of keyword terms with opposing counsel but instead wanted to use AI to meet deadlines and get to relevant documents faster?

Use Active learning without changing your workflows

IPRO now offers AI-powered active learning capabilities as a behind-the-scenes approach to eDiscovery.

As attorneys complete their normal eDiscovery processes, the tool runs in the background to give them insights into what documents might be relevant to their case and how many additional documents they still have to review. Attorneys can then choose to validate these results with the data they discovered in their traditional workflow (e.g. keyword terms).

Instead of requiring that AI be used by both parties, IPRO solutions simply augments their review. When they compare these results with what they previously found and begin to discover gaps and conflicts in their traditional review workflow, attorneys suddenly want to use AI more regularly in their processes to validate and get to the facts faster.

That’s what makes our new AI-powered solution more powerful now. It’s not asking attorneys to change how they practice; instead, it’s offering new insights based on the information it was given.

It finds conflicts in their reviews and allows them to validate their own methods so they can feel comfortable with trusting AI predictions and using the technology more. More importantly, it allows them to feel confident after multiple successful cases and be able to defend the technology when necessary.

Think upstream

Using AI to validate these upstream review processes can have a hugely beneficial effect on downstream eDiscovery processes.

Once attorneys fully trust the technology, they will be more apt to further train it by creating targeted reviews to be ready for a deposition — or just sample the data and assess information prior to a massive data collection effort — or even compare their facts against the data produced by opposing counsel.

Once you gain insights into key information, focusing on relevance rather than filtering data is a powerful method.

Every attorney should have an AI toolkit in their back pocket. It will arm them with information they didn’t know they had.

Trust in AI is everything

It all goes back to trusting the technology.

The general feedback from our active learning beta testers was that has been they didn’t want to introduce anything new into their processes but instead, they used our behind-the-scenes solution approach with a few of their cases and saw how quickly they got to evidence but more importantly, how quality was a major factor.

That feedback quickly changed to “wow, this tool can validate where we are, easily find conflicts, and allow us to quickly assess the data–and we don’t have to change our process.”

We would have never seen that kind of quick acceptance in the past because we had been taking the wrong approach all along.

NOW, I can’t wait to see even more adoption in the future.

Learn more about our active learning capabilities.