What we call 'machine lawyering'
This blog uses the term “machine lawyering” to refer to the ensemble of processes and questions that are arising from the application of information technology to law, legal transactions and regulation. It includes most of what is thought of under the rubrics ‘fintech’, ‘regtech’ and ‘lawtech’. From the point of view of practicing law and regulating activity, machine lawyering can be understood to take three concrete forms and provokes a number of related questions:
Form 1: automation. Machine learning is being applied to automate analysis of documents for specific items of interest, such as scanning documents received in anticipation of litigation or in the context of a corporate acquisition, as well as a first review of contracts. Blockchain technology is being used to automate the steps of relatively simple transactions into ‘smart contracts’ that allow performance by one party automatically to trigger counter-performance of the other without human intervention. Data retrieval and drafting systems, like that of IBM’s Ross, are automating the process of legal research – the initial gathering of relevant law in response to facts presented by a client.
Form 2: surveillance. At least since Mike Aitken developed SMARTS in the 1990s, data analytics have been used to monitor securities trades for market abuse. Now, better data flows and smarter analytics are allowing focused surveillance for compliance with nearly every type of law, from securities trading and funds transfers to meet market fairness and anti-money laundering rules, to pricing behavior for compliance with competition law.
Form 3: privatization. Distributed ledger technology (DLT) is often acclaimed for its decentralizing potential to liberate transactions from central authority. If the cryptography is robust enough to permit unrestricted access and the costs of hardware maintenance requirements are not too onerous, DLT could allow most transactions now recorded on registers held by a trusted central authority to be dispersed into open networks. This could effectively ‘privatize’ activity from the clearing and settlement of listed securities to the recording of deeds for real property.
Labor. The automation and surveillance activities of machine lawyering will have a significant impact on the current state of labor. Routine work young lawyers might once have used for early training, much of it now already in the hands of paralegals, will be passed to machines. As the capacity and skills of artificial intelligence to undertake legal research and draft memoranda of law increases, another large part of what young lawyers do will be automated. Regulators too will experience savings on labor, but a potential increase in activity could lead to more, rather than less, jobs for lawyers. Analytics applied to data flows from the regulated markets will allow regulators to root out abusive behavior at levels previously unheard of. While the automated data processing will decrease the need for analysts, it could well increase the need for prosecutors and other enforcement personnel.
Data protection and intellectual property. Two large, supplemental questions that arise from machine lawyering fall under the headings of data protection and intellectual property. With personal data being captured and processed in ever more intrusive and comprehensive ways, the protection of personal data is becoming as important to daily life as tort and criminal law have been for centuries. As machines increasingly produce analyses and work products, decisions on the ownership of such solutions will determine winners and losers in a market that could include nearly all complex legal solutions by 2050. A related issue is the regulation of how data processing algorithms are structured and to whom they should be disclosed for review and potential regulation. Particularly in the area of surveillance, as such algorithms make concrete decisions on whom and what falls under investigation (or at least initial review for investigation), algorithm architecture should become a focus of public and judicial scrutiny.
Machine lawyering will impact every aspect of the legal profession. Law is essentially information, so none of the physical restraints that impede automation in other areas will limit the reach of information processing in the law. While reasonable persons may disagree on whether this development is positive or negative, few will dispute that machine lawyering is here, and that it will change the practice of law and regulation as we have known it.
David Donald, Hong Kong