University of Johannesburg: A brave new world needs brave new laws that address artificial intelligence

When the University of Johannesburg hosted the first World Congress of Constitutional Law to be held in Africa, we reflected on the 17th-century philosopher John Locke, who argued that the law of nature compelled humans not to cause harm. We could not help but also realise that this obligation of not harming humans extends to artificial intelligence (AI) machines, which increasingly use human agency in decision-making.

Isaac Asimov extended John Locke’s observation by proposing the three laws of robots, which are that a robot: 1) will not harm a human; 2) will obey a human as long as this obedience does not lead to the violation of the first law, and 3) will protect itself as long as this protection does not lead to the violation of the first and second law.

Locke’s argument became the basis of the American Constitution, implemented in 1787, and countries throughout the world would draw on these principles. While this principle still underpins constitutionalism, in a rapidly changing age, transformational constitutionalism comes to the fore. To define this quite simply, transformational constitutionalism ushers in a change in social, legal, economic or political systems based on the values set by a constitutional framework.

We are certainly in a period of immense change. We are undergoing humanitarian crises ranging from extreme poverty, climate change, hunger and lingering and persisting inequities. The pandemic has exacerbated these crises, and the aftermath has been challenging to navigate. We also find ourselves amidst the Fourth Industrial Revolution (4IR), defined by significant disruption to technology, industries, and society, which calls for changes to policy and our approaches.

Betwixt and between this changing landscape, transformation continues to be an essential conversation for us to drive. Considering the Constitution as our means to navigate this terrain, it stands to reason that constitutionalism must reflect our society and the changes we undergo.

How do we govern in uncertain and rapidly shifting times? Thurgood Marshall, the first black Supreme Court Justice in the United States, once said “today’s Constitution is a realistic document of freedom only because of several corrective amendments. Those amendments speak to a sense of decency and fairness.”

What, then, is the interface between the law and these emerging technologies? For example, self-driving cars — in other words, cars that do not need a driver because they drive themselves — drive on our roads and, therefore, are subject to our rules and regulations, such as speed limits. The question, however, is who is responsible for the fine if the self-driving car runs a red light or exceeds the speed limit? Moreover, who is accountable in the case of an accident?

In Arizona, an Uber self-driving car killed a pedestrian. According to preliminary investigation reports, this self-driving car observed the passenger but went ahead and killed her. The backup driver was charged with involuntary manslaughter for the crime committed by the AI self-driving car.

The Department of Transport is formalising rules around self-driving vehicles in South Africa, but liability remains a talking point. Is the manufacturing company liable? Or the AI company? Or the human operator if present? Arguably, the flaw rests in the product. Our legislature needs to enact laws that will ensure that these self-driving cars and any intelligent machine in our factories operate based on an understanding of the principles of AI and its implications.

Another example is gene editing. AI and machine learning have emerged as tools that make gene editing initiatives more accurate, cheaper and more straightforward. For instance, Crispr-Cas9, a novel biotechnology derived from bacteria, can make precise changes to specific locations in the human genome. The potentials range from combatting genetic diseases (germline editing) to optimising human health.

Conversely, “optimising” the human race through eugenics, for example, has historically proven to be a tool to marginalise and eradicate people. There is a tightrope of morality we are walking. Our legislation needs to respond accordingly to protect people. How do we tap into the benefits of gene editing without risking violating human rights?

In the South African Journal of Science, Thaldar and co-authors argue that we need to tap into the potential of germline editing based on the principles derived from the Constitution. Genetic editing is expected to comply with the same laws and ethical requirements as all scientific research relating to human reproduction.

However, the authors argue that principles such as the regulation of the process, ensuring the well-established standards of safety and efficacy, measures for allowances of non-therapeutic uses of germline editing, respect for the individual’s reproductive autonomy and equal access should guide new regulations.

A third example is the use of facial recognition software. In the shift towards biometric systems, various ethical concerns have been raised. Research suggests that facial recognition has inherent biases. For example, police in the US use Idemia, which scans faces using algorithms. Yet, results from the National Institute of Standards and Technology have indicated that two of Idemia’s algorithms were pointedly more likely to confuse black women’s faces than white women’s faces, or black or white men. Where Idemia’s algorithms erroneously matched the faces of white women at an accuracy of one in 10,000, it incorrectly matched black women’s faces about once in 1,000 — so 10 times more frequently.

Then, there are concerns about surveillance and autonomy. In South Africa, the biometrics policy proposed in 2021, which suggested that the biometric information of all citizens and residents in the country be centralised into one database, was criticised based on privacy concerns and the potential for human rights infringements. Calls have been made to include clear guidelines for facial recognition software, big data and information matching that emphasise security safeguards for the ethical and responsible use of surveillance technologies.

Though these examples merely scratch the surface, the takeaway is that emerging technologies need to be aligned with existing human values. Constitutionalism remains a guiding force in this endeavour. It should be recalled that the Constitution of the Republic of South Africa, 1996 is the country’s supreme law. Furthermore, any law or conduct that is not consistent with the Constitution is invalid.

It can be taken as a given that more technology-related questions of law will arise for our courts to adjudicate. Thus, court officials (i.e., judges and magistrates, prosecutors and legal practitioners) must keep abreast of any developments regarding emerging technologies and how they interface with the law. Cutting-edge research by academics to assist the courts with legal questions on the interaction between AI and the law is more necessary than ever.

As Canadian Justice Michel Bastarache puts it, “… academics can predict the consequences of rules and legal decisions under hypothetical situations and possible future conditions… Academics are busy writing about decisions that have not even been written yet and are doing studies of issues that have not yet been addressed by courts… No principled approach to decision-making can ignore the contribution of academics.”

Another important point to note is that one cannot downplay the important role (to be) played by statutory bodies such as the South African Judicial Education Institute, which has been established to, among others, promote the accessibility and effectiveness of the court through continuing judicial education. Life-long learning with regard to the interface between the law and emerging technologies by court officials is absolutely necessary if they are to uphold the Constitution and dispense justice in a meaningful manner, particularly in the context of the 4IR.

Currently, we are in the infancy stage of regulation in the sphere of AI and the law. Realistically, to ensure that our legislation is in line with 4IR, Parliament will need to look at all our legislation and update it in line with the 4IR. This would require Members of Parliament and Cabinet to become 4IR and science literate.

Specifically, the creation of intellectual property rights stands out in this context because the very principle of a creative and knowledge economy implies the rapid production of new technologies, artefacts and processes for commercialisation and scale.

As Bhimrao Ramji Ambedkar, the former Minister of Law and Justice of India, once said, “the Constitution is not a mere lawyers’ document, it is a vehicle of life and its spirit is always the spirit of age.”