Forward-looking legal principles and regulation of predictive technologies

I gave a presentation on 'Forward-looking legal principles and regulation of predictive technologies' at the 2025 ALSA Conference on Law and Technology held at UNSW on 10 July 2025. The following is the abstract:

The use of predictive and prescriptive algorithms and AI systems by public and private actors is becoming more commonplace. While data analytics and AI have clear benefits, they are also vulnerable to errors, misuse and abuse. Well-known examples include the flawed automated system of the Robodebt scheme and questionable use of algorithms in criminal sentencing and bail decisions. Regulators, scholars and the public tend to be preoccupied with how AI and data analytics will impact or transform law. This presentation adopts the opposite approach and seeks to examine how future- and compliance-oriented legal principles can inform and accommodate the use of predictive and prescriptive technologies in regulation. The presentation though goes beyond the conventional approach of recommending universal or common legal principles and values that need to be considered such as transparency, accountability and privacy since these have already been addressed and are expressly contained in Australia’s AI Ethics Principles and the Council of Europe Framework Convention on AI. Instead, it delves into specific legal rules and concepts that are especially relevant to the issues of prediction and prescription. For instance, it first discusses the forward-looking dimension of law and how it anticipates and deals with future acts and behaviours. The analysis covers, among others, the statutory interpretation principle of the law ‘always speaking’ and the notion of technological neutrality in legislation and explains how these are relevant to the forecasting functions of data analytics and AI. Furthermore, the presentation examines the normative character of law and enforcement-related legal principles such as natural justice and how they are applicable to the decision-making goals of automated and algorithmic systems.

For more presentations, visit the Presentations page.