Two of the hottest terms in the technology of law today are “artificial intelligence” and “machine learning.” From the prose devoted to them, one might expect that they will launch a new era, with a potential for dramatic and fundamental change in law practice. One must, though, exercise caution. Artificial intelligence has been a promising technology for decades, predicted to arrive very soon. Both phrases may be misdirecting. Regardless of their advertising, these are computer programs and are subject to the present, and perhaps fundamental limits, of computation.
We should, however, understand and embrace the impact that computation has to empower and enhance the practice of law, and be prepared for the developments to come.
Artificial intelligence can be misdirecting in the sense that the full extent of what we consider intelligence—including wisdom, judgment, experience, purpose, emotion, empathy, awareness—is well beyond the power and methods of today’s technologies. The learning in machine learning also implies many of the characteristics that we ascribe to intelligence.
Computation, though, pervades our lives and work, a development that has occurred during the lives and work of today’s senior practice leaders. In their childhood, or just before, computers were people who performed computation, as portrayed for example in Hidden Figures. Then our tools, work and play, employed little or no computation, whether we were writing a document, sending a letter, storing correspondence, notes, and other content in a file, recording dictation or other sounds, taking photographs or movies, making a phone call, searching for authorities or precedents, or retrieving contents of a file. The entertainment we heard and watched and the cars we drove and the maps that guided our travels used little or no computation. Recall the AAA Trip Tik, a booklet of printed strip maps created especially for vacation travel. It still exists and is now an app. Every one of these activities and tasks today incorporates computation, and most now cannot function without it.
I focus on computation rather than all of the facets of technology. Computation, by computer or a system of computers—whether they are presented as a computer or incorporated in another tool— consists of a set of programmed steps on data that execute or present a result. The steps depend on arithmetic and logic. A computer’s logic also depends on arithmetic. Computation can incorporate very large numbers of operations and very large quantities of data in a very short time to a high degree of precision. In all of those respects, today, computation exceeds the capabilities of unaided humans, including lawyers. These capabilities amplify and extend what human lawyers and staff can do. They cannot perform all of the process of thought, of the elements of intelligence listed above.
Luminance, a UK-based legal technology vendor, asserts that its artificial intelligence programming and machine learning capabilities allow it to “understand language by context and content.” Put to work to assist in document review and due diligence, according to Luminance’s website, the program can search large collections of documents very quickly and identify information: “By pairing machine learning algorithms and advanced mathematical techniques with deep domain expertise, [our product] finds the clauses and documents within the data room that deviate from the norm.” The search capabilities and methods accomplished by these products are important, and represent significant advances in the computation that aids the practice of law.
The program can learn because the choices a user makes feedback into the results. We are increasingly familiar with the results of this kind of programming when we shop on Amazon and other sites, and our purchasing history—even our browsing history—is used to direct us to similar products or to other products purchased or investigated by others who have viewed items similar to those we viewed.
Neither Amazon nor the due diligence system understand in the sense we ascribe to humans. The computations they perform allow them to gather, display, and highlight matching results, or where important, anomalous results. Understanding still requires human (read, lawyer) participation.
Leading law practices will implement teams that fully empower their work with computation, where the computation performs tasks at a speed and depth that cannot be matched by humans alone, together with the interpretation, evaluation, expression, and perspective that remain the domain of the human lawyers. Together computation and law practice has and will have a power that neither of them can alone.