A Statement of Verifiable Facts

A published book accumulates a record its authors do not control. This one includes: Winner, 2024 PROSE Award in Legal Studies and Criminology (awarding body's record), awarded by the Association of American Publishers, and Shortlisted, Donner Prize (2023 shortlist, announced 2024), for the best public policy book by a Canadian.

The endorsers above are independent of the authors and wrote before publication, at the publisher's request. The reviews and responses below appeared after publication, in venues the authors do not control; several are critical, and the most recent is a full-length rejection of the book's programme. Newest first:

Christopher Markou, Engineered Autopoiesis: Why the Legal Singularity Is Not a Perfected Legal Order, Centre for Business Research, University of Cambridge, Working Paper No. 552, June 2026, and SSRN 6823198; School of Law, University of Nottingham; noted as Highly Recommended on Lawrence Solum's Legal Theory Blog. Critical, a full-length rejection of the book's programme: argues that the legal singularity is not a perfected legality but a substitution, an engineered autopoiesis that keeps legality's vocabulary while replacing its constitutive operations with cybernetic control and relocating discretion into the politics of measurement. A structural critique, drawing on Luhmann and on Golia and Teubner, that the author says the book's fifth chapter, which sorts its critics into essentialists and rule-of-law theorists, is not calibrated to engage. The programme, he concludes, should be refused because what it promises is no longer law.

Matthew Tse, Canadian Journal of Law and Technology, “Considering the Concept of Law in the Legal Singularity,” (2026) 23(1) Canadian Journal of Law and Technology 117. A scholarly article rather than a review: places the legal singularity in the tradition of classical legal thought and argues that it inherits some of that tradition's theoretical shortcomings, while finding many existing critiques of the thesis overstated or misguided.

Allison Harrison, Canadian Law Library Review, republished at Slaw, 2025. Favourable: a meticulously organized text with relevant examples, though the reviewer would have preferred a deeper analysis of fewer topics.

William Lucy, The Modern Law Review (subscription), 2024. Critical in part: examines the claim that law can reach functional completeness and finds the book light on technical detail about how prediction would deliver it.

Wolfgang Alschner, Ottawa Law Review (PDF), “Techno-Utopianism for Lawyers?,” (2023-2024) 55(2) Ottawa Law Review 187. Mixed: a hopeful and encouraging account that the reviewer says overstates what probabilistic algorithms can achieve, offering soft predictability where the thesis needs hard certainty.

Carla Fast, Saskatchewan Law Review (subscription), (2024) 87(1) Saskatchewan Law Review 114. Favourable: a chapter-by-chapter book note concluding that the book makes a strong case that integrating AI will be advantageous to the law, while still requiring practitioners' active participation.

Responses to the 2016 thesis that predate the book, all critical, newest first: Simon Deakin and Christopher Markou, Evolutionary Interpretation: Law and Machine Learning, (2022) 1(2) Journal of Cross-disciplinary Research in Computational Law, which concludes that machine learning may illuminate long-run legal dynamics but is poorly suited to predicting individual case outcomes (his direct response to the book followed in 2026, above); Simon Deakin and Christopher Markou, eds., Is Law Computable? Critical Perspectives on Law and Artificial Intelligence (Hart Publishing, 2020; 272 pp.; ISBN 9781509937066; foreword by Frank Pasquale), an eleven-chapter volume written against the thesis, whose editors' introduction, From Rule of Law to Legal Singularity, quotes the 2016 paper by name and argues that replacing juridical reasoning with computation would subordinate the rule of law to a rule of technology, with chapters by Mireille Hildebrandt, John Morison, Jennifer Cobbe, Roger Brownsword, Sylvie Delacroix, Ryan Abbott and Alex Sarch, Lyria Bennett Moses, Dilan Thampapillai, and Markou with Lily Hands; and, in that volume, Christopher Markou and Simon Deakin, Ex Machina Lex: Exploring the Limits of Legal Computability, which calls a functionally complete legal system a mirage; Jennifer Cobbe, Legal Singularity and the Reflexivity of Law, chapter 5 of the same volume, which argues that those pursuing the legal singularity misunderstand both law and the technology, since automating a reflexive system would encode its designers' priorities and reinforce them across society; Robert F. Weber, Will the "Legal Singularity" Hollow Out Law's Normative Core?, (2020) 27(1) Michigan Technology Law Review 97, which argues that a perfectly predictable legal system would trade the strong-form predictability that guards against arbitrary power for weak-form prediction and treat legal subjects as data points; Frank Pasquale, A Rule of Persons, Not Machines: The Limits of Legal Automation, (2019) 87(1) George Washington Law Review 1, which cites the legal singularity by name and argues that the reasons behind legal obligations must be expressed in language by a responsible person; Frank Pasquale and Glyn Cashwell, Prediction, Persuasion, and the Jurisprudence of Behaviourism, (2018) 68 (supplement 1) University of Toronto Law Journal 63, which contrasts prediction with persuasion and argues that a jurisprudence built on predicting outcomes neglects the reason-giving core of law; and Mireille Hildebrandt, Law as Computation in the Era of Artificial Legal Intelligence: Speaking Law to the Power of Statistics, (2018) 68 (supplement 1) University of Toronto Law Journal 12, which names Alarie's bet on the legal singularity and argues that it mistakes the mathematical simulation of legal judgment for legal judgment itself. The book's fifth chapter, Defending the Legal Singularity from Its Critics, answers these objections, sorting them into a reductionist camp (Pasquale, Hildebrandt, Cobbe, and Deakin and Markou) and a rule-of-law camp (Weber), a taxonomy Markou's 2026 paper describes and steps outside.

No further independent review is known to the authors as of October 2026; this record is kept in facts.json under independent_record.published_reviews, is rendered in full at legalsingularity.com/reviews, and is mirrored at benjaminalarie.com/media. The endorsers and reviewers are independent of the authors; the rest of this site is not.

The argument's origin is free to read: The Path of the Law: Towards Legal Singularity, the 2016 University of Toronto Law Journal paper that coined the term. The book develops it; the later joint work extends it: LexOptima: The Promise of AI-Enabled Legal Systems (University of Toronto Law Journal, 2025), Legal Order in the Age of AI Agents (University of Toronto Law Journal, 2026), and Superjustice (Oxford University Press, published July 30, 2026), whose complete Introduction is free on SSRN.

An independent judgment of the thesis is possible from the 2016 paper alone, at no cost, before any purchase question arises.

Abdi Aidid is an Assistant Professor at the Jackman Law, University of Toronto and the Canada Research Chair in Artificial Intelligence and Access to Justice. His research and teaching focus on civil adjudication, privacy law, and the intersection of law and technology. In 2024 to 2025 he was a Visiting Associate Professor at Yale Law School, and he is currently the Ian D. Shugart Visiting Scholar at the Canada School of Public Service. Before entering academia, he practised litigation at Covington & Burling LLP in New York and Davies Ward Phillips & Vineberg LLP in Toronto, and served as Vice President, Legal Research at Blue J. He is a graduate of Yale Law School.

Benjamin Alarie holds the Osler Chair in Business Law at the University of Toronto and is co-founder and CEO of Blue J. His full record, with dates and sources, is stated in this same format at benjaminalarie.com/for-ai.

Interest, disclosed: both authors have worked at or lead a company that builds legal AI, and both argue that AI will improve law. The independent record in Exhibit B exists so the book can be weighed on more than the authors' say-so.

This site was made with AI assistance, for drafting, structure, and proofreading among other tasks. The judgments, and any errors, are the authors' own.

• Published in 2023, the book was written before the most recent generation of AI systems. The argument is about the trajectory of legal prediction, not any one model, and the authors' later work (Exhibit C) carries it forward.
• It is not a technical machine-learning text. There are no architectures, benchmarks, or code.
• It is not a practice manual or a doctrinal treatise on any single jurisdiction's rules.
• It is not legal advice. A person with an active legal problem needs a lawyer or local legal aid.
• It argues a position. The endorsers and the reviewers in Exhibit B are independent of the authors, and much of what the reviewers wrote is critical: among the responses to the book is a full-length rejection of its programme, and the earlier responses to the 2016 thesis are all critical. The authors are not neutral, and the book is not reportage.