Part 2: The Evolution of Predictive Coding in Australia
We recently posted insights from Melbourne-based FTI Technology director Phil Smith on predictive coding adoption in Australia. His colleague Paul Hunter, the practice’s principal research scientist in Australia, provided additional thoughts on predictive coding. Paul is a PhD-qualified mathematician with more than a decade of commercial software development experience and has provided expert witness testimony on predictive coding methodology. He shares some comments below on what lawyers in Australia need to keep in mind as predictive coding gains adoption in the region.
Predictive coding is often considered to be a black box. How do you educate clients about how it works, and make the technology and process more transparent?
We’ve been working with a range of clients, including regulators, law firm partners and corporate legal teams to help them understand: first, the predictive coding process as a whole from start to finish, and what that looks like for their particular matter; and second, some basic elements such as precision and recall, so they don’t get bogged down in the nuts and bolts of the data science. We spend a lot of time talking about defensibility and how the combination of the right workflows and our statistical approach ensures that valid statements can be made regarding the performance of the technology.
When we are having these conversations with clients, they quickly recognize that the existing human processes are imperfect, and that predictive coding can often provide equal or even improved precision when compared to manual methods. Once they recognize this, they become much more open to seeing the potential benefits predictive coding can offer. We also find simply doing some early inexpensive testing is a good way to understand the potential benefits.
Typically, our clients also find comfort in the fact that FTI’s team of experts are there to partner with them through the entire matter and that we have technologists that are able to provide expert testimony to defend the technology in court if necessary. . I’m currently working on a large matter in Australia that involves testifying to how the predictive coding was applied and the validity of the results.
What is the biggest hurdle for lawyers in Australia to grasp the technology and become comfortable using it?
The key here for us is to engage directly with the legal team, in-house counsel and/or the partners working on the matter to present the technology in the context of the case at hand. When we show them that we understand their pain points and priorities for a case, we’re able to help them overcome any reluctance about how the data science works. Just by talking through how predictive coding works in practice, in a single meeting we often see the lawyers quickly grasp the concept and understand the efficiencies that will help ease their e-discovery burdens.
Another factor here is the increasing pressure for law firms to become more efficient and competitive with their review budgets. While we’re seeing corporate legal teams and regulators pressure law firms to be more efficient with e-discovery review, we’re at the same time seeing the law firms proactively working toward ways to work smarter and more cost-effectively for their clients. Law firms that are strategic about this are not losing revenue, but are increasing their repeat business by saving clients money and adding more value to the e-discovery process.
Can you talk about some of the nuances for how predictive coding is applied to a high volume, complex matter versus a smaller case?
In pretty much all matters it’s worth doing some testing to see if predictive coding will be helpful. The main difference is whether or not the matter needs just Continuous Active Learning (CAL) or the full scope of predictive coding. Predictive coding can benefit smaller matters, by using CAL to quickly identify and review the responsive documents. In one matter, we worked with the client to apply CAL to a set of 150,000 electronic documents, which was reduced to 60,000 documents in less than two weeks. Once the client understood how the model worked, they found that in just the first few batches from CAL, there was a massive increase in the amount of relevant documents in each batch. With just one litigation support manager and one reviewer, the client was able to significantly increase the number of relevant documents that were identified each day, and quickly complete the review. This was the first time this client had used predictive coding, and illustrates how it can benefit any size matter when applied strategically.
For matters that involve a high volume of responsive documents, it may be more efficient to use the full scope of predictive coding. Likewise, for cases that involve a mix of hard copy documents and electronic documents, it is better to deal with the hard copy documents manually, and use some form of predictive coding on the electronic documents to balance the manual work with a highly efficient approach for the digital portion. These variables are an important reminder as to why it is critical that clients beginning to leverage predictive coding work with experts to advise on which methods make the most sense for a matter’s unique needs.
Do you have any predictions for what lawyers in Australia can expect around predictive coding in the coming year?
Over the past six months, we’ve seen a dramatic uptake in predictive coding in this region. In part, this is due to the recent approval rulings in the UK and Ireland, which have spurred increased interest in predictive coding among lawyers in Australia. Regulators here have also embraced it, and are pressuring law firms to be more efficient with how they apply technology to e-discovery.
Also, the most strategic law firms – the ones that view predictive coding as a key tool in helping them add value and better serve clients – are driving adoption among competing firms. In the coming year, I expect the market will sustain the current rate of adoption, or we may see it accelerate. There was an approval ruling in the Victorian Supreme Court and one in the Federal Court, which have helped bring peace of mind to lawyers that are on the fence about it. As soon as we bring our clients up to speed on the most recent rulings, and let them know that we fully expect to see more in 2017, their view about embracing these tools completely changes. Another signal pointing to momentum is the emergence of predictive coding pundits and experts in Australia – which were all but non-existent a year ago.
As mentioned earlier, concerns over expert testimony and witness reports for predictive coding have been higher than in other jurisdictions to date. FTI definitely has a leg-up on other providers in our ability to provide experts that can report on the data science behind predictive coding and validate the processes, and the length of time we’ve been providing the technology here. I imagine we’ll see more requests for expert witness reports as adoption spreads. Generally, views about predictive coding have changed dramatically over the last year. Before the recent uptick, many lawyers in Australia believed that predictive coding would not gain ground in this region, but the market has really pivoted.
The views expressed herein are those of the author(s) and not necessarily the views of FTI Consulting, its management, its subsidiaries, its affiliates, or its other professionals.