[By Abhinav Gupta]
The author is a student at the National Law University, Jodhpur.
Introduction
While discussing his book, 21 lessons for the 21st Century, Yuval Noah Harari points out that humans face existential crisis due to technological disruption. In his other book, Homo Deus: A brief history of tomorrow, he highlights the good and bad of artificial intelligence (AI). He believes that AI holds the potential to exterminate mankind. The emergence of cases where algorithms and AI have replaced humans has made this prophecy more increasingly daunting.
Algorithms and AI are revolutionizing the way businesses operate. The disruptive effects of such technology have been felt across various business functions. These emerging technologies have the potential to create corporations that are completely autonomous and run by algorithms. At the same time, they can be used to just assist in increasing the efficacy of business operations. Business decisions require comprehending a variety of data and AI has proved its capability to process complex data to reach conclusions. The use of AI in boardrooms can help directors and benefit stakeholders by complementing them, if not substituting, in performing their functions. In this article, the author discusses how AI has the capability to revolutionize the functions of directors and what the operational and legal hurdles are in employing AI at the highest level in corporations.
Artificial Intelligence
AI has the ability to process huge amounts of data. While human intelligence can process only the seemingly important and related data, AI can process unrelated data as well, which might have a bearing on the decision. There are three kinds of AI, differentiated on basis of decision-making rights allocated to such AI, namely, assisted, augmented and autonomous. Under the assisted AI, the tech merely assists in the process of decision making and does not take decisions itself. The decision-making power continues to rest with the human. Augmented AI shares decision rights with humans and both learn from each other. On the other hand, autonomous AI completely replaces the human and operates independently of human intervention to take over all the decision rights. This distinction between different kinds of AI can be implemented in corporations in order to develop a robust technical environment.
Using AI in Corporate Functioning
The ability of AI, big data, and machine learning can be exploited to assist corporations in taking strategic decisions, managing risks and ensuring compliance. Moreover, humans are often faced by their cognitive biases which prevent them from considering certain relevant information or flip side of issues. AI, unlike humans, is free of these biases which can greatly affect the functioning of a corporation.
AI would help directors in exercising ‘independent judgment’ and become appreciative of various views as each suggestion by AI will be based upon concrete data. AI can be used to channel contrarian views based on such data and reduce the occurrence of ‘groupthink’. Groupthink refers to a situation where an individual tends to agree with the viewpoint of the majority in order to form a consensus, irrespective of the validity or correctness of such viewpoint. Hence, directors will be able to convey dissent in boardrooms which is essential in order to ensure that decisions taken by the board are in the best interests of all the stakeholders rather than just the directors or a select group.
AI can also be used for the selection of board members. With more and more information available about directors regarding their qualifications, past experiences, AI should be able to process this data to ascertain the future performance of the candidates in light of the objectives and future plans of the company. It would be easier for AI to comply with the law regarding the qualifications of directors. For instance, Section 149(6)(a) of the Companies Act, 2013 (the Act) provides that independent directors should have the relevant experience and expertise. Moreover, they should not have any pecuniary relationship with the company. Naturally, this would entail going through past transactions of the candidate as well as the company. An AI, which has all such data of transactions, would be much better equipped to determine if the candidate possesses such experience and expertise and whether they have any pecuniary relationship with the company, ultimately helping in compliance. The availability of data allows AI to foresee trends and at the same time handle the data of past and present, efficiently. Thus, AI can assist in the early detection of non-compliance, allowing the company to mitigate penalties and punishment associated with such non-compliance.
It has been the approach of corporate law scholars that boards must be monitored in order to uphold the interests of shareholders and prevent self-serving directors from putting themselves before the corporation. The Indian regulator has conformed to such an approach by keeping checks on directors and providing for their duties (see Section 152; Section 166; Section 169; Section 171 of the Act). By keeping a record of all the transactions undertaken by the directors, AI allows keeping a tab on the functioning of the board to ensure compliance with the law. Be it reporting related party transactions (see Section 188 of the Act) or whether directors are complying with their duties under Section 166 of the Act or Schedule IV of the Act.
Moreover, AI helps in handling the agency problem. The agency problem refers to the conflict of intentions of a principal and agent. Where the principal expects the agent to work in furtherance of his best interests, the agent would have certain interests of his own and might prioritise them over the principal’s interests. However, AI does not have any agenda of its own. It would operate on the basis of the available information and how it is employed by the directors. AI would act to the best of its capability in the best interests of the stakeholders rather than pursuing its own agenda.
Directors authority and Duty to delegate to AI
After affirming that AI has the capacity to make informed decisions, one must understand whether the directors have the authority to delegate their decision-making rights to AI. Neither the Act nor any case law directly addresses the question of whether directors can delegate decision rights to AI. However, under the Act directors do have the power to delegate certain tasks.
Few cases dealt with the question of directors power to delegate and therefore, analogies can be drawn to apply those principles to the present issue. In Bhagwati Prasad v. Shiroman Sugar Mills Ltd., it was held that the articles of association of a company provide the director’s authority to delegate their powers. In light of this, if we look at Model Articles of Association under the Act, it provides the directors with the power to delegate. Article 71(i) states that “the Board may, subject to the provisions of the Act, delegate any of its powers to committees consisting of such member or members of its body as it thinks fit.”[1]
If such provision is incorporated in the Articles of association of the company, it provides sufficient power to directors to delegate. However, the question remains if they can delegate it to AI in particular. AI cannot be considered to be a committee or a member, which implies that the board cannot delegate its power to AI but simply employ it for their assistance. This would mean that the decision-making rights would remain with the directors and AI can merely help to make such decisions.
However, the object of such provisions is to allow the directors to delegate certain tasks in order to ease the burden and work in an organized manner. Keeping in mind this object, if we give the provision a broader interpretation, then we can justify saying that powers can indeed be delegated to AI.
In light of this, the companies that seek to incorporate AI into the boardrooms and delegate essential tasks to it must provide for the same in the articles of association in order to avoid conflicts at a later stage. An explicit mention of such authority to delegate to AI would not leave any iota of doubt. Moreover, considering the Act does not anywhere prohibit any such delegation, incorporation in Articles of Association would be sufficient and not contrary to the law.
At the same time, we can consider if directors have the ‘duty’ to delegate to AI. Directors of a company have to exercise their duties with due and reasonable care, skill and diligence and exercise independent judgment (see Section 166 of the Act). The Act thus requires directors to act on an informed basis. Considering that AI is considered to be better at decision making as it can process huge chunks of data, the duty to act on an informed basis may transform into a duty to rely on all available technological advancements such as AI. However, until the use of AI becomes the norm for corporations, it cannot be said that directors were not diligent in case they did not resort to the assistance of AI.
Can AI replace human directors?
The question that still looms large is whether AI can replace human board members. Researchers believe that AI will be capable to replace humans on board for two reasons; first, the tasks performed by directors are become increasingly complex and second, AI is better equipped to perform functions of the board as compared to humans. A Hong-Kong based firm Deep Knowledge ventures appointed an algorithm named Validating Investment Tool for Advancing Life Sciences (VITAL) as their board member. Vital was appointed to make decisions regarding investments by the firm using its data processing capabilities and information that humans are incapable to comprehend. Another AI, Alicia T. was appointed as a member of the board by a Finnish tech company, Tieto.
If we look at the Indian law on appointment of directors i.e. Section 149 of the Act, it states that every company should have a board of directors consisting of ‘individuals’. A minimum of three directors is required in a public company and two directors in the case of a private company. It is debatable if AI can be considered an individual and if a legal personality can be ascribed to such technology. In light of this, once the statutes declare or the court of law decides that AI indeed is a legal personality, then under Indian law, AI might replace human directors on board. In absence of any such provision, it can be reasonably concluded that the current regime does not allow a company to have AI as directors on its board.
At the same time, AI as a director also raises issues with respect to pinning liability in cases of default. Directors liability has been an issue of contention and they are liable for various defaults under the Act. In case a default made by AI, it will cause the courts of law to scratch their heads regarding who to pin the liability on. Will the directors who chose the algorithms and AI be held liable? Can courts order corporations to stop the usage of such AI? It remains to be seen how the courts in India tackle these issues.
Conclusion
The way the human mind is wired is significantly different from how AI functions. For this reason, it cannot be said that the regulations formed keeping in mind human intelligence is sufficient to tackle issues posed by AI. AI’s capability should be exploited in order to keep in line with the dynamic technological environment. Involving AI in the board room does not mean we can completely do away with human directors. AI can help the company take decisions where such decision is data-driven rather than requiring a ‘human touch. The boards in the future need to be equipped with the knowledge of algorithms and AI to chothee an AI best for their organisation. The boards’ composition will change accordingly to accommodate business experts rather than accounting and monitoring experts.
[1] Article 71, Articles of Association of a Company Limited by Shares, Table F, Schedule 1, Companies Act, 2013.