Artificial intelligence and healthcare in South Africa : ethical and legal challenges

Abstract

We find ourselves in an era of intelligent machines as technology pushes the boundaries towards the fourth industrial revolution. AI is already here in disruptive narrow forms, cutting across the globe and leaving almost no sector untouched. With numerous industries such as the healthcare sector already adopting AI technology, we continue to see increased benefits for our society as a whole. AI is developing at a rapid pace and is already capable of undertaking human tasks, and even more effectively and at a lower cost, creating novel challenges for recognised regulatory frameworks and posing serious ethical questions. AI systems are used by professional healthcare workers to diagnose problems or to take decisions on the method of treatment for patients. As such, these AI systems can be regarded as either products or services through which healthcare is delivered to patients as the consumers. The thesis therefore considers how AI ought to be regulated in South Africa, elaborating on the ethico-legal concepts that can be applicable to the branches of the law with emphasis on the healthcare sector. Two main groups of parties could be held accountable under civil law in the event of injury sustained in the application of AI systems, comprising of the producers and the end-users of the systems. In South Africa, where damage ensues due to defective AI systems in healthcare, the injured party would have a claim against the producers or professional end-users of the systems, either in delict based on negligence where the plaintiff carries the duty of establishing that the duty of care of the reasonable expert or professional has been breached or in terms of product liability as codified in the Consumer Protection Act (CPA). In law of delict in South Africa, it makes no difference whether AI systems are classified as a product or a service, since both will be actionable in terms of the Aquilian action or non-patrimonial loss when damage is unlawfully caused. Similarly, this distinction does not apply under the CPA. Thus, in delict the injured party is allowed greater flexibility in terms of pursuing an action either under the negligence principles or under the strict product liability principles under the CPA as opposed to the other jurisdictions. Notably quality assurance software standards such as the ISO series can be applied in establishing negligence in AI liability and such standards can motivate for the software and computer programming industry to be conferred with a professional status in South Africa to provide a benchmark that registered software engineers and technicians must meet as globally recognised professional standards. The thesis also recognises the values that need to be protected in South Africa when considering the ethical implications emanating from the use of AI in healthcare and considers a working ethical framework based on international ethical guidelines that have already been developed for AI.