Md Ferdows Hossen
Lawyer| Researcher| Author| and recipient of the Top 25 IP Rising Star Award: Talks about #IP & AI Governance
From the big bang to big data, every revolution has left some deep marks on human history and its concept of law, be it for better or worse. The new values and norms that were created throughout the Industrial Revolution,[1] revolutionized and reorganized the order of society across the world.[2] Likewise, the tech revolution, more specifically, the advent of artificial intelligence (AI)[3], has transformed virtually every aspect of human life and society. For example, Uber already has self-driving cars on the streets that are more cost-minimizing on the one hand and drive more safely than human drivers on the other hand, although the trolley dilemma has not left it yet.[4]
Today, AI outperforms humans in a diverse array of human lives[5], such as algorithms buy and sell nearly half of all trades by volume in the US and European stock markets.[6] AI-powered killing machines like drones and Advanced Targeting and Lethality Automated Systems (ATLAS) have reshaped the idea of war, along with the international humanitarian law that confronts its viability.[7] Furthermore, AI is already in the loop of taking life-or-death decisions in the medical world while robot judges are deciding the soft cases. We will probably not need lawyers anymore for compliance decisions, avoiding the legal liability. Interestingly, just before this century began, an artificially intelligent machine defeated the best grandmaster in chess.[8] Nearly a decade after that, an AI deadly challenged the grandmasters of the television trivia show Jeopardy![9] Last but not least, they can guess what consumers most like to have at the restaurants.[10]
As for the AIs’ omnipresence, tech history has already witnessed a number of great breakthroughs. Google’s Deep Mind and Apple’s Siri are two significant inventions among them. On the other hand, in late 2022, most tech giants, following Open AI’s Chat GPT, launched their generative AIs.[11] It is now commonly predicted that there will be some super artificial intelligence soon that will be able to feel emotions, anger, and act just like humans, though some scientists see it with deep doubt and consider it a myth.[12]. As of now, the presence of AI has transformed many sectors, most significantly law and its concepts, such as the pattern of contracts,[13] replacing the rules and standards with “Dynamic Rules”[14] and “Micro-directives”[15].
Historically, the separation thesis, a centuries-old legal doctrine, stands as a demarcating fence between the two most influential jurisprudential schools: legal positivism and natural law. The legal positivists prerequisite that the law must be descriptive, morality-neutral, and derived from an ‘is’; by contrast, the natural lawyers reject the idea of “is” and claim that the law must be evaluative, morality-dependent, and derived from an “ought” (A. Marmor: 2017). In other words, law, in the legal positivists’ view, is always source-based; contrarily, natural lawyers count law as merit-based. Let us take an example to understand the separation thesis clearly. If one narrates that “X is Y”, the statement is true if X, indeed, is Y, and false if X is not Y. Again, if one asserts that “X ought to be Y”, one’s statement “X is not Y” does not render his statement false unless one says “X ought not to be Y”.[16] Academically, innumerable research papers, articles, and books have already been written in diagnosis of the separation thesis since it emerged in the literature. But interestingly, this jurisprudence-old debate has not yet ended.
Evidently, AI has reformed the foundational conceptions of the law. Today, the traditional rules and standards are being consolidated into micro-directives, a vast catalogue of ex-ante context-specific rules. While translating those rules into simple commands that are communicated to the regulated actors as situations demand, this changes the whole traditional law making-process. To be more comprehensive about the case, let us borrow an example of traffic lights:
“In a world of rules and standards, a legislature hoping to optimize safety and travel time could enact a rule (a sixty miles-per-hour speed limit) or a standard (“drive reasonably”). With micro-directives, however, the law looks quite different. The legislature merely states its goal. Machines then design the law as a vast catalog of context-specific rules to optimize that goal. From this catalog, a specific micro-directive is selected and communicated to a particular driver (perhaps on a dashboard display) as a precise speed for the specific conditions she faces. For example, a micro-directive might provide a speed limit of 51.2 miles per hour for a particular driver with twelve years of experience on a rainy Tuesday at 3:27 p.m. The legislation remains constant, but the micro-directive updates as quickly as conditions change”.[17]
With the example above, one cannot deny that legislators are now exempted from trading off between rules and standards, a resource-costing job. On the other hand, making ex ante rules by lawmakers and adjudicating ex post standards by the judges are taken over by artificially intelligent machines. All that may be left for legislators and judges to do with rules and standards is to set algorithms with some set of goals into artificially intelligent machines to act accordingly.
AI and its associated deep learning models, on the other hand, may push humans to the edge of total collapse and cataclysm. To prevent the likelihood of this disaster, policymakers and industry leaders agreed upon adopting some set of ethical codes to compel the machines to be aligned with.
Documentarily, every organization has a self-styled code of ethics to architect and sculpt artificially intelligent machines, and when they design AIs based on those codes of ethics, machines accordingly end up providing merit-based and morality-dependent rules to the actors.
Extant research has mainly focused on the rules governing AIs, the rights and liabilities of AIs, crimes and offenses relating to AIs, the personhood of AIs, and AI agency, while some other research papers have shown that the making of rules and standards, the concept of law, and jurisprudence have remarkably been affected by AI-driven norms. For example, Brian Sheppard, in Warming Up to Inscrutability: How Technology Could Challenge Our Concept of Law (2017), examined how Hart’s reasonable officials, Raz’s legitimate authority, and Dworkin’s interpretive theory are facing challenges in the age of technology. Yueh-Hsuan Weng and Takashi Izumo, in their Natural Law and its Implications for AI Governance (2019), proposed what ways natural law could guide and help to mitigate the conflicts between the soft and hard laws regulating and governing AIs.
N.B: for the full version of this paper:
[1] Industrial Revolution, HISTORY, (Oct. 10, 2022, 2:40 PM), https://www.history.com/topics/industrial-revolution/industrial-revolution.
[2] Douglas W. Allen & Yoram Barzel, The Evolution of Criminal Law and Police during the Pre-Modern Era, 27, Journal of Law, Economics, & Organization, 540–67, (2011).
[3] AI includes Robots, Machine Learning, Deep Learning, Neoral Networks, Algorithms, and Super AI (under process).
[4] Sean Gerrish, How Smart Machines Think 56 (Cambridge, MA : MIT Press, 2018).
[5] Martin Ford, The Rise Of The Robot: Technology And The Threat Of A Jobless Future (1st Ed., Basic Books, 2015).
[6] Simon Chesterman, We, The Robots? 21 (1st ed., Cambridge University Press, 2021)
[7] Id. at 45
[8] See generally Bruce Pandolfini, Kasparov And Deep Blue: The Historic Chess Match Between Man And Machine (1997).
[9] John Markoff, Computer Wins on ‘Jeopardy!’: Trivial, It’s Not, N.Y. TIMES (Oct. 23, 2022, 2:30 AM), https://www.nytimes.com/2011/02/17/science/17jeopardy-watson.html.
[10] thomas w. Miller, modeling techniques in predictive analytics: business problems and solutions (1st Ed., Pearson Education Inc., 2014)
[11] Mark Beccue, AWS, Microsoft, and Google Cloud: Tying Up LLMs, THE FUTURUM GROUP, (Oct. 23, 2023, 2:30 AM), https://futurumgroup.com/insights/aws-microsoft-and-google-cloud-tying-up-llms/.
[12] See generally erik j. Larson, the myth of artificial intelligence: why computers cannot think the way we do?, (1st ed., The Belknap Press of Harvard University Press, 2021)
[13] Anthony J. Casey & Anthony Niblett, Self-Driving Contracts, 43:1, The Journal of Corporate Law (2017) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2927459.
[14] John O. McGinnis & Steven Wasick, Law’s Algorithm 66 FLA. L. REV. (2014)
[15] If Computers Wrote Laws: Decisions Handed Down By Data, THE ECONOMIST, (Oct. 19, 2022, 2:00 AM) https://worldif.economist.com/article/12133/decisions-handed-down-data.
[16] I took up this example after being inspired by A. Marmor’s Draft Article ‘What Is Law and What Counts As Law? The Separation Thesis In Context’.
[17] Anthony J Casey & Anthony Niblett, The Death of Rules and Standards, 92 :4 Indiana Law Journal 1401, 1403, (2017).