Artificial intelligence (AI) is becoming the key technology of the future. AI is the ability of a machine to simulate human behaviours such as reasoning, planning and creativity, particularly by learning algorithms. Companies are increasingly turning to AI to develop more automated, personalized and “customer-oriented” services. AI also opens up new opportunities to enhance and facilitate the detection and mitigation of risk and fraud and to enable better regulatory compliance. However, its use is not without risk, as illustrated by Deloitte in one of its research reports:
• Quality, quantity and relevance of the data used. The results of AI systems are dependent on the quality and quantity of the data. If the datasets used to develop the algorithms contain biases, the algorithm generated is likely to reflect these biases as well, or even amplify them.
• The opacity of operation (black box in the context of AI). Contrary to previous generations of AI, where the systems made decisions that were very clear and established by humans, the new generations will rely on very complex statistical methods, based on thousands of parameters. All these factors make the final decision difficult to interpret, or even impossible to explain by humans.
• Possible dysfunctions. Algorithms do not have the capacity for conceptual understanding and common sense that humans have, and which are necessary to assess radically new situations. As highlighted in the latest reports from the World Economic Forum, the subject of artificial intelligence is reaching a turning point.
In the short term, it seems important that the development of artificial intelligence meets the minimum criteria in terms of governance, ethics and risk management. Again according to Deloitte,2 this reflection should focus on proving the reliability of the algorithms used (from the point of view of their internal and external verification), the intelligibility of the models and the interactions between humans and intelligent algorithms. It is therefore crucial that the mandate of the Corporate Governance Committee be reviewed in order to integrate this reflection and to develop a code for the use of artificial intelligence in order to assure shareholders and interested parties (stakeholders) that its development and use are carried out by placing humans at the heart of the machine and by guaranteeing the accuracy, security and confidentiality of the data that feeds it, and by regulating the algorithms so that they integrate diversity and overcome biases in decision making, in particular.