AI in the boardroom

TECHNOLOGIES AND BIG DATA ARE INCREASINGLY FUNDAMENTAL ELEMENTS OF DECISION MAKING PROCESSES. TO AVOID ABUSES, ERRORS AND INEQUALITIES, MORE COMPETENCIES AND SOFT LAW AT THE GOVERNANCE LEVEL, NOT LESS AI

by Maria Lilla' Montagnani e Maria Lucia Passador, associate professor and academic fellow at Department of Law

During the pandemic and in the aftermath, AI is revealing itself as an essential tool in dealing with some of the intricacies that have emerged. In fact, by processing a huge amount of data, AI allows the identification of important correlations, it enables to perform key studies for the containment of the current health crisis by tracking the mobility of people living in the areas most affected by the pandemic or by quickly recognizing the symptoms characterizing the onset of the virus. In this way, doctors can focus on truly infected patients, immediately quarantine them and give them the most appropriate medical care.
 
With the emerging potential of AI becoming increasingly clear, the reliance on this technology as a means not only of recovery but, more broadly, of transformation is being strengthened at all levels, including that of corporate management. Companies that succeed in leveraging the use of AI in the most appropriate way are not only the most likely to more effectively overcome the crisis but also to succeed in the digital economy, which requires investments in digital skills development and training, cyber risk protection, high-speed data connectivity and high-tech solutions. In particular, AI can be a fundamental tool for management, given its “data digesting” capacity, i.e. the ability to intake huge amount of data, extract relevant information and make suggestions on the path forward. Depending on the AI applications that directors adopt, the technology can be used to make corporate reporting more efficient, to improve a company's legal compliance, and ultimately to make more informed strategic decisions.
 
However, while it is undeniable that AI has many extremely useful applications both in the social and business fields, the risks that could potentially arise from its employment are equally undeniable. Indeed, as of now, there are remarkable cases in which automated decision-making - implemented through AI - resulted in discriminatory decisions in a wide range of areas, from employment law to the allocation of subsidies, from adjudication processes to criminal law. In the case where AI is used by managers, there is some concern about the risk of them relying on the AI’s indications without being able to understand its reasoning, thus becoming mere executors – since understanding the AI reasoning is a fairly complicated task. In fact, one of the main issues of AI in all sectors is that of its “opacity”, which makes impossible to understand how AI achieves its outcome. This in turn raises another question about who is going to be liable when an AI-driven decision harms the company and its shareholders. If the AI is unexplainable, and directors slavishly follow its suggestions, are they still to be considered responsible for the result of their conduct?
 
To profitably bring AI into the boardroom, the use of AI applications must be governed in a way that minimizes the risks that can result from their use. First, this means developing the skills to deal with AI applications by increasing diversity within boards and including those with technology expertise. Second, this entails the adoption of tech committees – or, if a company already has a risk committee, the expansion of its purposes to incorporate AI oversight. Such a committee should be vested with functions like assessing the proper use of information disseminated, identifying key business risks, reviewing reports periodically submitted about the AI system, monitoring the adequacy and effectiveness of the chosen IA system, and assessing that precise area of expertise, i.e., "technology," at its fullest.
 
In conclusion, best practices for the use of AI by boards need to be developed, and this should be done at the soft law level, through corporate governance codes for their greater pliability, flexibility and role as a laboratory for solutions to be tested.