Tailoring \u3ci\u3eex Machina\u3c/i\u3e: Perspectives on \u3ci\u3ePersonalized Law\u3c/i\u3e

In their book Personalized Law: Different Rules for Different People, Omri Ben-Shahar and Ariel Porat propose a radical approach to lawmaking: using of big data and artificial intelligence to tailor legal dictates to the individual histories and characteristics of persons they affect. This essay critically discusses that proposal. It first examines normative differences among the Ben-Shahar and Porat’s proposals for personalizing laws. There are important differences, for example, between using big data and artificial intelligence to tailor how a private legal power can be exercised to the capacities and interests of the power-holder and imposing different speed limits on different drivers depending on their personal histories. The desirability of personalization depends both on the type of legal rule (duty, power, privilege, etc.) and on the type of traits personalization attends to (abilities, propensities, preferences, etc.). The essay then identifies a few reasons to worry about using big data and artificial intelligence to tailor legal dictates to the individuals they affect. This approach to lawmaking would make it more difficult to detect misfeasance and malfeasance in the legislative processes. And lawmaking that issues individualized commands rather than general rules and that provides no explanation for why different people receive different treatment poses a threat to the rule of law. Lastly, the essay argues that Ben-Shahar and Porat’s proposal relies on a false picture of practical reasoning. On their proposal, legislators would need to specify in advance both a law’s goals and the relative weights of those goals, after which they would turn matters over to computer scientists, data managers, and statisticians. This assumes a degree of practical omniscience lawmakers are unlikely to have. Successful practical reasoning requires iterative and reflective inquiry into both ends and means together. Successful lawmaking involves revising, supplementing, and sometimes abandoning preliminary goals as lawmakers reason through the possible means of achieving them. Algorithmic lawmaking is not compatible with such practical reasoning about ends