How Will Machine Learning Impact Legal?aderantuser
In my last article I posited that the humanistic functions of a lawyer are not under any immediate threat from artificial intelligence, but that automation technology will play a steadily increasing role in how businesses are run and how software is designed.
That said, advances in artificial intelligence (AI) and data analysis still appear poised to dramatically affect your practice. The human method of legal reasoning is not that far removed from certain aspects of predictive analytics and holistic machine learning, but is limited by the time we have available. Many legal experts now believe that AI will play an increasingly important role in the legal industry both in legal practice and in practice management—and to some extent this is already happening.
Writing for Slate back in 2011, Farhad Manjoo noted that “The legal industry is one of the few remaining outposts of the corporate world whose operations are dictated mainly by human experience. Basic questions that anyone would want to know before committing to a million-dollar case – How likely is it that I’ll win? How good are my lawyers? Should I settle? – can’t be answered with certainty.”
The current reality of the legal industry is obviously still far removed from robotic lawyers. But big data and predictive analytics are nevertheless having a significant impact. In a recent post on ILTAnet.org, Deborah Dobson put forward several key areas where firms are already using big data, even around dynamically using social data to assist in jury selection.
So what is predictive analytics and machine learning? Predictive analytics has been around for a long time, simply being the analysis of complex historic data to forecast future trends; something most of us do through our BI products today. But with significantly more data available and a lot of it unstructured (such as images and documents), it is difficult to analyse and find the patterns that we need, especially as the questions themselves are not always clear. This is where machine learning fits in. Using AI algorithms we are able to, in parallel, analyse immense quantities of both structured and unstructured data (so called “big data”) and use these algorithms to identify patterns that traditional data structures and queries would never expose.
This trend is already very evident in e-discovery, with vendors and firms heavily investing in more pattern-based analysis and filtering, creating a whole new genre of tools and approaches to case preparation and processing, in addition to requiring new skills in interpreting and driving this new approach. This isn’t without its challenges—a great blog post by Johanna Scholtes explores how hard it is to interpret unstructured text in e-discovery.
Studies have shown that humans are generally not very good at predicting the future, especially when the timeframe is extended to any significant length. Our various biases and motivations tend to cloud our ability to impartially interpret all the variables in such a task. Computers don’t have that problem although, as Hawking and Bostrom suppose, they are likely to outstrip us in intelligence within 100 years (or 15 years according to Google), we hope they don’t rush to having motivations to go with it.
Whatever the case, we should open our eyes to the possibilities inherent in this technology. As people, we look to the past to help us predict the future. But we also tend to focus on the obvious patterns rather than hidden ones. While there is a clear and steady growth in this area of law, have we analysed for themes in the most common questions being asked by our clients or our social contacts and looked for opportunities around these?
Big data and AI will almost certainly affect your practice in the coming years. You may not be hiring robot lawyers, but you will likely be looking for answers in that ocean, or competing with people who have gone diving for them.