Artificial Intelligence - normally understood as the simulation of human intelligence processes by machines, especially computer systems. These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions) and selfcorrection. The research and test-cases on AI have been in existence for over half a century now - as early as in 1940s and 1950s. Right from Turing’s test, the game AI to the Dartmouth conference, the idea of AI was conceived decades ago. It is only now that the discussion has gathered momentum. Why should you care? In the world that we live in, we are both the users and contributors of the ‘intelligence’ of the not so artificial bots around us. Siri, Amazon e-commerce, Netflix, Google Maps, Spotify are only a few examples of the leading players that use AI solutions. As both a user and a contributor, is important to know the dynamics of an AI system.
Our newsletter this month, touches upon the attribution of liability and privacy in case of AI and assesses the extant principles of law in UAE in this regard. We also discuss the Ethical AI Toolkit launched by Smart Dubai earlier this year.
Who is liable if anything goes wrong in case of an AI is a wide spectrum of debate. Can responsibility for damage caused by AI be attributed to someone? Think of this - in case of a basic AI solution, the minimum number of supply chain players would include:
The most common of the actionable wrongs which one could fathom based on the above supply chain would amongst others be - product liability - contractual liability; and - tortious liability for breach of statutory duty, wrongful harm, and negligence.
As opposed to the rules of attribution in the conventional legal systems, where a legal person is separate from a natural person, or where the companies/ corporations hold separate personalities, AI is likely to disrupt traditional allocations of liability in the AI supply chains. The ‘rules of attribution’ in case of an AI would go through myriad of assessments and possibilities including the possibility of attribution of a separate legal personality or agency on an AI in itself. While most jurisdictions have shied away from conferring a legal personality on AI , it’s only a while before this debate heats up. For instance, Article (291) of Federal Law No. (5) of 1985 On the Civil Transactions Law of the United Arab Emirates (“Civil Code”) reads “When several persons are responsible for a prejudicial act, each one of them is responsible for his share in it and the judge may decide to allot the liability equally between them or consider them jointly and severally responsible.” Article 282 on the other hand reads “The author of any tort, even if not discerning, shall be bound to repair the prejudice.” It’ll therefore be interesting to see the possibilities of attribution of liability and the obligation of making good the loss or compensating for the damages in case of an AI. Who shall qualify an author for the purposes of Article 282, in case of an AI system will be event more interesting to see. In case of basic contractual liability, as per Article 141 of the Civil Code, a contract may only come into existence when there is an agreement between the two parties to the contract concerning the essential elements of the obligation. Article 151 of the Civil Code also states that if a person makes a commitment on his own and for his account, then he shall be bound by the provisions of it to the exclusion of other persons. Assume the case of a simplest subscription based online service, which uses AI solutions back end. As a subscriber, you’ll mostly have a clear offer and acceptance leading to the formation of a contract (told you to read the T&Cs!). If there were to be any breach of the said contract, which could be attributed to the ‘bad judgement’ of AI solution, you can only imagine the level of dispute from the perspective of a user and the supply chain participants mentioned above. And then think of the dynamics of most basic principles of contracting like the doctrine of privity, specific performance, strict liability and damages OR, look for an advisor to guide you!
Privacy, by definition, implies the power to isolate oneself. In most countries, right to privacy is a fundamental right. With AI’s ability to collect, analyze and store huge data, the right to privacy has been affected immensely. Some may see the technology disrupting an individual’s rights, while others may visualize legal consequences. In UAE, the general right to privacy arises out of the Constitution which provides that “freedom of communication by post, telegraph or other means of communication and the secrecy thereof shall be guaranteed in accordance with the law. The Civil Code under Art 31 allows the individual to seek compensation for any “unlawful infringement” of his or her privacy. Acknowledging the importance of privacy, the Penal Code under Article 378 provides that anyone who misuses or discloses any private or personal data of an individual is punishable by fine and imprisonment. AI and machine learning features may cause the processing of personal data in a way that may lead to loss of control on the data by the data subjects, and may therefore, attract one of these provisions. Additionally, Article 2(1) of Federal DecreeLaw no. (5) of 2012 on Combating Cybercrimes provides that whoever gains access to a website, an electronic information system, computer network or information technology means without authorization or in excess of authorization or unlawfully remains therein shall be punished with an imprisonment or a fine.
The bespoke federal law concerning IT communications in the healthcare sector indicate the government’s commitment to ensure privacy protection. Smart Dubai Ethical Toolkit From Uber’s self-driving car fiasco to Amazon’s gender biased recruitment tool, the concerns of ethical concerns persist and must be addressed before an AI tool is deployed. An argument that has mostly received, consensus among the industry experts is that AI needs to reflect the societal ethics and values. There is also a consensus that AI does not have a mental states and so far AI does not have personhood. But what is possible is that cognitive behavioral capabilities become indistinguishable from that of a human in certain directions. In case of an AI ‘intentionality is coded’ and so is ‘Mens Rea’. Striving to fix some of the problems in the AI ecosystem, Smart Dubai’s Ethical AI Toolkit has been created to provide practical help across a city ecosystem. It supports industry, academia and individuals in understanding how AI systems can be used responsibly. It consists of principles and guidelines, and a selfassessment tool for developers to assess their platforms. The major guiding principles of the toolkit are fairness, transparency, accountability and explainability.
An argument that has mostly received, consensus among the industry experts is that AI needs to reflect the societal ethics and values. There is also a consensus that AI does not have a mental status and so far AI does not have personhood. But what is possible is that cognitive behavioral capabilities of AI become indistinguishable from that of a human in certain directions. Afterall, in case of an AI: ‘intentionality is coded’ and so is ‘Mens Rea’!
To stay updated,
subscribe to our newsletter