News

Justice Hercules

The role of judges in judicial review is to uphold the law in cases where its meaning is contested. They must figure out what the law is, what it permits, and forbids. They must be impartial, with an unyielding loyalty towards the law, to think, to weigh logically, to exercise objective judgment. They should also declare the reasons for their judgement, and that might also be subject to appeal.

In his book Thinking Fast and Slow Daniel Kahneman captures triggers of human thinking processes and lists some flaws in the thinking process of a group of judges being studied (parole judges in Israel). According to this study the judges were more prone to grant a parole after the meal break, with the approval spikes reaching 65% after each meal, whereas on average only 35% parole requests got approved[1].

Ronald Dworkin aware that consistency by Judges cannot easily be achieved, in his book, Law’s Empire[2], conjures Justice Hercules. A super human being, who alone could discover and apply flawless theory of  adjudication, by knowing all the laws, past cases, applicable theories and thus which would be able to justify the use of principles in ‘hard cases’ by constructing the best theory of law possible. It is somewhat symbolic that for the task at hand Dworkin used a fictitious demigod figure, with unparalleled strength and who, according to Greek mythology, was the only demigod to accomplish the mythical 12 labors set out for him.

Should we be surprised that judges are human, thus not flawless, or infallible, or that they may suffer from glucose depression? Do we need an infallible Herculean justice or flawless lawyers in an unerring legal system for justice to be served and our rights to be preserved? Is this even possible? Some argue that technology can answer these questions for us and could also be the solution for all our legal problems. A multitude of applications in the LegalTech industry are opening up opportunities of AI for the legal profession. Machine-learning algorithms and natural processing language are already used to assist in delivering court judgments abroad. Applications such as IBM’s Ross, dubbed the first AI lawyer, allegedly also answers legal questions posed by users with astonishing accuracy, even providing citations and suggestions for further reading. A Law firm in the USA has also recently announced has announced that they are employing IBM’s AI Ross to handle their bankruptcy practice[3].

There are also legal futurists like Benjamin Alarie who predict that AI will bring legal singularity (a hypothetical point where computational intelligence and decision-making capabilities exceed those of human lawyers, judges and other decision-makers) borrowing the term “singularity” from Vernor Vinge, and it will replace the existing legal regime through an automatic interactive process between the algorithmic systems of governments, lawyers, corporations, and solving all legal woes in the process[4].

Max Tegmark, a physicist, and co-founder of the Future of Life Institute, to this end also opined that:

Since the legal process can be abstractly viewed as computation, inputting information about evidence and laws and outputting a decision, some scholars dream of fully automating it with robo judges: AI systems that tirelessly apply the same high legal standards to every judgment without succumbing to human errors such as bias, fatigue or lack of the latest knowledge[5].

A robo judge from this angle would not suffer from the glucose depression identified by Daniel Kahneman in his book Thinking Fast and Slow. Despite the recent advancements in technology however, AI is still somewhat narrow in its application, and from a technological perspective seriously doubt we have reached a juncture where an AI expert machine or a cluster of such machines is able right now to replace a whole legal system or our judges. Having said this, I cannot exclude the possibility that trough the attainment of Artificial General Intelligence and more connected expert machines or other new technologies, we might have software artifacts which in the future can outperform intricate legal work exclusive to the domain of lawyers or judges.

Given the nature of the singularity it is likely that a new and evolved legal system would replace the existing one with the following:

  • Algorithmic artifacts replacing the existing structures in law.
  • Digitalized legal and administrative eco system with an abundance of digital data to guide automated decision making.
  • Seamless enforcement of legal rights through deterministic measures.
  • AI, judges able to take decisions based on this data as well as in situations that require abstract reasoning and where there is insufficient data upon which to base a decision.

On the latter point, Judge Posner does not exclude the inception of AI judges a priori albeit he playfully posited that “The judicial mentality would be of little interest if judges did nothing more than apply clear rules of law created by legislators, administrative agencies, the framers of constitutions, and other extrajudicial sources (including commercial custom) to facts that judges and juries determined without bias or preconceptions. Judges would be well on the road to being superseded by digitized artificial intelligence programs. I do not know why originalists and other legalists are not AI enthusiasts[6]

This hypothetical futuristic scenario I tried to capture above, would surely reduce human bias, error, doubt, and inefficiency, since coded and automated laws can be directly enforced by the non-empathic and efficient Herculean Ai judges. Thus, in essence, this will transform law and the metering of legal provisions and justice into an algorithmic process, through efficient intelligent artifacts. Will this really benefit us humans and place us in a better position?

I believe that other considerations aside from technology would need to be assessed here to avoid hubris of technological prowess. The legal system is not only text or coded laws and something static, as much as a musical masterpiece is not only musical notes on a piece of paper but finds its beauty and meaning in artistic interpretation. The legal system is a dynamic and living organ affected by political, economic, and socio-cultural factors that influence its nature, application as well as its evolution. Should technological prowess really take us there, I am still not clear how the new algorithmic system will capture and preserve the complex and dynamic nature of this living organ as a social institution, a construct of and within society. Will the existing social order be preserved, will the existing actors prevail in this new regime, and will there be scope to incorporate the assumptions, priorities, and values of these actors which currently form part and feed the existing legal system?

This vital organ is reflective in nature, nourished from the principles of predictability and universality, (Locke & Rousseau), essential components as well for the rule of law. The rule of law is not confined exclusively to a set of laws or rights which for the sake of argument could be fully coded into software, but also clips back to the democratic system, which bring these laws and rights alive. How will the proposed legal singularity effect or change this? What will the effects be?

As part of the system within our rule of law, we have decentralised and separated powers and roles trough the distinct institutions, the legislative, executive and judiciary, to make sure that the nice words written on a paper or even potentially coded into software (called laws and rights), are protected and effective. The envisaged legal singularity, if it materializes, might replace the humans behind the existing institutions or else nullify the existing separation of powers by recentralising partially or totally the formulation, adoption, interpretation, and enforcement of laws through software artifacts. Will the evolved legal singularity follow a design and operational process as the defunct legal system it replaced, where possibly oracles or other AI’s or humans will check and balance the process and eventual outputs? As Hart observed, this is critical to ‘preserve the sense that the certification of something as legally valid is not conclusive of the question of obedience, and that, however great the aura of majesty or authority which the official system may have, its demands must in the end be submitted to a moral scrutiny’.[7]

When Deep Blue (the computer chess player) defeated the then reigning grand master, Garry Kasparov in 1997, it did not replicate the way he played, but rather uses the strengths of the machine (fast, precise, infallible memory) to play better than a human. It is not farfetched to think that the same could happen to law. Perhaps the use of technology soon would like wise introduce new approaches and techniques which are not the ones used by humans due to our limitations, but which would give better results[8]. Several legal scholars have also remarked that rather than challenging the existing shortcomings in the legal system, singularity here promises to recode these and automate their logic. This in turn according to them, would affirm and reproduce their designers’ and users’ subjective assumptions and ideological priorities as well as their accompanying societal contexts and structures through feedback loops. One would need to ponder here what role and rights will citizens have with legal singularity in this new environment.

There are already strong views on the use of AI in courts, especially by the CCBE[9] ( Council of Bars and Law Societies in Europe) which attest that such AI systems would amongst others run afoul of the existing rule of law if they:

  1. Use of data and elements that have not been the subject of adversarial debate.
  2. Exploit conclusions that have not been obtained through the reasoning of the judge.
  3. Lack of transparency of the process, since it will be difficult what should be attributable to the judge or the machine.
  4. Undermine the principle of impartiality due to the impossibility of neutralising and knowing the biases of system designers.
  5. Breach of the principle of explicability due to the exitance of results that are beyond huma reasoning and cannot be traced.
  6. Remove or dent the notion of level playing field (equality of arms)

Ugo Pagallo and Quattrocolo Serena also focused on the use of AI and automated evidence gathering, raising issues with Article 8 (right to a private life and also right to a fair hearing)[10] In the case of Wisconsin vs Eric Looms[11] the use the automated COMPAS (Correctional Offender Management Profiling for Alternative Sanction) was like wise attacked on lack of  principles of due process (Fifth and Fourteenth Amendments).

In Europe aside from rules on due process, we already have laws like the General Data Protection Regulation, which would prohibit a decision taken by an AI judge based solely on automated processing, without  consent as well as Article 11 of Directive 2016/680 on Data Protection in Criminal Matters is even stricter on this issue as it prohibits them in toto, unless authorised by the respective member states. Added to the latter even the new draft regulation on AI aims to capture the Herculean AI Judge under its remit and impose human oversight. Maintaining oversight and resilience to avoid unintended consequences is duly warranted, however we also need to understand that simply putting AI genie in a panopticon will not address the issues we have at hand.

Whilst I do not believe that we need to have an infallible Herculean justice or flawless lawyers in an unerring legal system for justice to be served, the legal sector can benefit from technological innovation. Like other areas transiting to a Society 5.0[12] it is already being affected by technological innovation and will reshape in ways similar to what Richard Susskind premised in his books. Civilization is likewise transmuting and whilst it might need new rights, it will always need a relevant legal system, which can function as a core and vital organ, reflective in nature, with the capacity to evolve and emancipate itself to the new reality, alea iacta est.

Article by Dr Ian Gauci, Managing Partner GTG Advocates and Afilexion Alliance. Special thanks are due to a dear friend and eminent scholar Professor Gordon Pace who has entertained long discussions and debates on the subject with the author and contributed with detailed analysis and suggestions.

Disclaimer: This article is not intended to impart legal advice and readers are asked to seek verification of statements made before acting on them.


[1] [1] Kahneman, Daniel. Thinking, Fast and Slow (p. 44). Penguin Books Ltd.

[2] Law’s Empire Belknap Press of Harvard University Press; 1st edition (January 1, 1986)

[3] . https://futurism.com/artificially-intelligent-lawyer-ross-hired-first-official-law-firm

[4] B Alarie, ‘The Path of the Law: Toward Legal Singularity’ (2016) 66 University of Toronto Law Journal 443, 445, papers.ssrn.com/sol3/papers.cfm?abstract_id=2767835

[5] M Tegmark, Life 3.0: Being Human in the Age of Artificial Intelligence (Allen Lane, 2017) pg 105.

[6] (RA Posner, How Judges Think (Harvard University Press, 2008)

[7] HL Hart The concept of law pg 21

[8] For further reading on this, there is also the case of chess champion Marion Tinsley who refused to play against the machine, Chinook see, “One jump ahead : computer perfection at checkers” by Jonathan Schaeffer.

[9]https://www.ccbe.eu/fileadmin/speciality_distribution/public/documents/IT_LAW/ITL_Position_papers/EN_ITL_20200605_CCBE-Response-to-the-consultation-regarding-the-European-Commission-s-White-Paper-on-AI.pdf

[10] Pagallo, Ugo; Quattrocolo, Serena “The impact of AI on criminal law, and its two fold procedures” [2018] ELECD 1416; in Barfield, Woodrow; Pagallo, Ugo (eds), “Research Handbook on the Law of Artificial Intelligence” (Edward Elgar Publishing, 2018

[11] https://epic.org/algorithmic-transparency/crim-justice/Wisc-Brief.pdf

[12] https://www.japan.go.jp/abenomics/_userdata/abenomics/pdf/society_5.0.pdf