Legal Technology Barriers — Understanding Language and Exercising Judgement

ContractStandards Blog
ContractStandards
Published in
4 min readJun 3, 2016

Originally posted in The Legal Executive Institute, September 24, 2015

There is a growing buzz around artificial intelligence (AI) and the law. My next few posts will explore emerging technologies and different approaches, and how they apply to emerging contract automation technology.

Technology is undoubtedly becoming ever more capable. But, there is also significant skepticism — and fear — over the rise of technology and its potential threat to the status quo. Doubters point to the limitations of technology, on the one hand, and the unique characteristics of human capacity, on the other, which, skeptics assert, cannot be replicated by an algorithm.

Addressing such concerns serves as a good starting point for an examination of advanced legal technology. Over the years of designing and developing legal technology, I have become acutely aware of the core challenges, namely emulating human expert capability of cognition and judgement. Human experts have unique skills to understand language and apply experience to counsel their clients. Surely, the subtle nuances of communication and situational circumstances must be beyond the capacity of a machine.

There are two important barriers to the development of this type of legal technology — the ability of computers to use judgement and to understand language.

Judgement

The first barrier is expert judgement: our ability to apply experience and counsel. One of the most widely publicized critiques of computer capacity is found in Hubert L. Dreyfus’s book What Computers Still Can’t Do: A Critique of Artificial Reason (the MIT Press, 1972). Prof. Dreyfus’s appraisal centers on the assumption that machines cannot emulate all aspects of human intelligence. However, this is not the goal of researchers in the field of AI. Each company and research institution is working on a particular task, sometimes very narrow tasks, rather than attempting to construct a generalized intelligence. Over the years, the machines have undertaken increasingly complex tasks.

The evolution of technology mirrors the development of human understanding. In exercising judgement, we must first find the relevant source material (or precedents).

In law, the exercise of judgement comes in a wide range of situations from basic to advanced. The evolution of technology mirrors the development of human understanding. In exercising judgement, we must first find the relevant source material (or precedents). Next from this material, we must identify the relevant facts and issues that may be dispositive of the matter at hand. And, finally, we must predict the optimal result or outcome. Legal technology is, in fact, well-advanced in this continuum.

In transactional practice, we can search and find relevant precedents, such as an asset purchase agreement. Technology can analyze the source materials and identify the relevant clause elements as assess whether each term in the agreement is standard, optional or deal specific. Finally, by examining the principal deal terms in a large body of precedent, technology will be able to predict the best deal structure and propose a draft agreement.

Language

The other great challenge is the difficulty of understanding human communications, both written and oral. Language is subtle and nuanced; the meaning of a sentence can be changed with negation through the use of “not” or subtly adjusted with rich vocabulary. In lawyerly terms, we can point to the complexity of understanding the differences between in sui generis and ejusdem generis. In terms of natural language understanding, parsers face the challenges of polysemy, synonymy, hyponymy and hypernymy and how the meaning of words are changed by their context.

Indeed, compared to the exercise of judgment, understanding language is the greater challenge, and technology has a long way to go. However, the pace of innovation is accelerating. To see how powerful the technology has become: just speak to your mobile phone and ask a question such as: “OK Google, what is the statute of limitations for assault in Illinois.”

In law, there are two principal factors that serve to mitigate the challenge. First, there is a general trend towards simplification of language. For example, average number of words in a sentence has fallen over the centuries. (Although it has to be said that legal documents and scientific manuals are still rife with wordiness.) This simplification of the language has made the written word more capable of being “read” by computer.

Second, globalization and increased commerce fuels the need for greater objectivity in our business relations and has driven the need for machine-readable, or -auditable contracts.

Conclusion

The need for speed, objectivity, and qualification in business relationships is pushing the capabilities of technology, while at the same time legal process management is guiding human behavior towards more predictable and repetitive forms. The two challenges I described above — judgement and understanding language — are but two barriers we are striving to overcome.

As we push technology (and technology pushes us) toward solutions, we may ultimately determine that there are no right answers for all situations. We will discover — with the help of computers — a quantum theory of law that will allow us to assess probabilities rather than certainties.

--

--

ContractStandards Blog
ContractStandards

A free library of standard contracts, checklists, and clauses built with sophisticated analytics and simple legal language. www.contractstandards.com