Governance by Numbers When Numbers Don’t Lie: The Cybernetic Path of Law Towards Legal Singularity
M Tinnirello and T Lozano (eds.) The Global Politics of Artificial Intelligence (CRC Press 2019, Forthcoming)
Posted: 12 Mar 2019 Last revised: 18 Apr 2019
Date Written: February 20, 2019
With the Fourth Industrial Revolution upon us, Big Data has become the organising principle of modernity, and mathematical quantification its preferred mode of expression. One consequence of this is that computational rationality—most familiar in the form use of algorithmic decision-making (ADM) systems—has supplanted communicative rationality and the notion of a discursive public sphere. In Governance by Numbers, French legal scholar Alain Supiot thus identifies a corresponding shift in constitutional legal orders from government by the state to ‘governance by numbers’ where mathematical quantification has become the basis for a new normative legal order. Some even suggest that the accumulation of more data and methods of better inference might culminate in a so-called "legal singularity"—an idea derived from the 'technological singularity' popularised by Ray Kurzweil.
In this world of a ‘legal singularity’ law is said to operate in a perpetual state of equilibrium between facts and norms. With AI vastly exceeding the capabilities of human lawyers and judges, both become superfluous to legal adjudication and process. Advanced as an answer to the incompleteness and contingency of the law, the project for the legal singularity is implicitly a proposal for eliminating juridical reasoning as a basis for dispute resolution and the allocation of powers, rights and responsibilities. As such it risks undermining one of the principal institutions of a democratic-liberal order. While there are many questions about whether so-called 'AI Judges' could replicate the cognitive domain of human lawyers and judges, the more important question is whether they should.
this paper suggests that the ideology at the heart of ‘governance by numbers’ and the 'legal singularity' is ‘dataism’—the belief that ‘everything that can be measured should be measured, and that mathematics and statistics provide an ontologically superior way of understanding and acting in the world. In his Homo Deus, historian Yuval Noah Harari declares ‘dataism’ a new religion succeeding humanism and bearing a central tenant 'that the universe consists of data flows, and the value of any phenomenon or entity is determined by its contribution to data processing.' The belief that data will allow for more reliable predictions further into the future turns on the empiricist notion that given sufficient volume data will ultimately speak for itself. This is 'governance by numbers' with a prevailing dataist orthodoxy that ‘numbers don’t lie’.
Dataism might, however, be reaching its apotheosis with the Smart City. There ubiquitous computation, ambient intelligence and predictive analytics will allow for increasingly granular decisions about resource procurement and allocation but also total informational awareness over the behaviours of inhabitants. Thus, in its aspirational form the Smart City establishes the sufficient conditions for realising the 'legal singularity' if not the necessary ones. This paper thus explores how the Smart City paradigm challenges constitutional legal orders by operationalising a cybernetic understanding of society as a series ‘of self-adjusting interacting units automatically responding to signal inputs and feedback, as programmed by computer algorithms.’
This paper traces the history of efforts to mathematise law beginning with 17th century German polymath Gottfried Leibniz, through the formalist legal axiomatics of Christopher Columbus Langdell, the legal realism of Oliver Wendell Holmes to the advent of logical-AI Legal Expert Systems in the 1970s-80s. Despite the Legal Expert Systems project failing due to an underestimation of applying the axiomatic method to law and legal reasoning, a new generation of so-called LegalTech applications leveraging modern AI techniques hopes to succeed at what logical-AI systems could not. It links attempts to mathematise law to the digital ontology of pancomputionalism—the view ‘our external physical reality is a mathematical structure’ and that the physical universe is essentially mathematical in nature. As a consequence, the hypothesis that ‘everything is mathematics’ legitimises the assumption that computation can not only ‘solve intelligence’ but following the Church-Turing Thesis; any problem solvable with computation—including legal problems and reasoning.
As the AI colonisation of law continues, largely unimpeded by regulation and incentivised by public-private partnerships, concerns about meaning and intelligibility of computational systems, and whether they can actually do what they claim, take a backseat or are considered superfluous to the causal structure of social reality. As a consequence, the extension of ADM to more contexts and calls for ‘AI Judges’ are bolstered by the uncritical push for ‘smart’ solutions to societal problems. However, this comes at the expense of engaging with whether legal reasoning can be reduced to mere functional capabilities regarding extraordinary information gathering, speed, memory, recall and even the ability to distinguish and disambiguate relevant legal rules. The paper argues that the effect of this is to substitute a new 'rule of technology' for the 'rule of law'.
Keywords: Artificial Intelligence, Smart City, Big Data, Datism, Scientism, Socio-Legal Theory, Systems Theory, Surveillance Capitalism
JEL Classification: K40, H10, L50, O33, O35, Z18
Suggested Citation: Suggested Citation