How Pioneering Boards Are Using AI

In 2014 Hong Kong–based Deep Knowledge Ventures formally appointed an algorithm to its board of directors, giving it voting power on the VC firm’s investment decisions. At the time, the appointment was seen as a gimmick. The algorithm simply analyzed quantitative data according to parameters chosen by humans to produce a base recommendation that the board members could debate. It was useful because it was faster than a human data analyst, but few observers believed that virtual board members would become commonplace.

Fast-forward a decade, and you might think that the remarkable advances in machine learning would have changed minds about the value of AI in the boardroom. But largely, that hasn’t been the case. From June to September of 2024, the authors conducted several focus groups with more than 50 board chairs, vice chairs, and committee chairs from public and private companies in Europe, Asia, and North America, including ASM, Lazard, Nestlé, Novo Nordisk, Randstad, Sandoz, and Shell. When they asked participants to rank the importance of various issues claiming their attention, they placed AI relatively low—well beneath issues such as global context and interaction with the CEO or significant shareholders. Most people reported using AI occasionally for personal needs but never or rarely for fulfilling their professional roles.

But in light of recent developments in AI technologies, and with virtual humans becoming commonplace business interfaces, the idea of an AI bot that engages in boardroom discussions doesn’t feel like a gimmick anymore. And while most of the focused group participants were skeptical of AI’s use for their board work, a significant number did share positive experiences of using AI.

AI can contribute to boards’ work in three ways.

Assisting individual board members. Directors are part-timers; typically they meet just four times a year and often serve on multiple boards. Yet they make key decisions for the organizations on whose boards they serve.  nonexecutive directors are detached from the operations and have limited time. As a result, they struggle to absorb the large quantities of information available. AI can help. A properly trained LLM such as ChatGPT can analyze large volumes of data to discover relevant patterns and trends that might not be apparent through the manual analysis directors themselves perform. It can continuously monitor various risks and provide early warnings, enabling proactive risk management. What’s more, it can condense all that information into easy-to-read formats, reducing the time and effort required for board members to process it.

Supplying the whole board with better information.  Most directors are fans of scenario planning. 

nevertheless, few boards practicing real scenario planning and analysis. Most chairs justify the omission by saying that the exercise would be too complex and that the board’s resources are limited.  AI can fix this, as it needs much less time than a team of human experts would to identify and assess potential changes in underlying variables and estimate their likely effect on the company’s value.  Some boards are using AI to give them a reality check on their decisions.  AI can also analyze board processes.

Joining the board. The logical next step is to have AI actively participate in boardroom discussions, and this is starting to happen. In 2024 United Arab Emirates’ largest publicly traded entity, IHC, appointed Aiden Insight, a virtual human, as a “board observer.” Aiden Insight doesn’t have voting rights, but its participation in IHC board discussions goes into the official minutes.  Aiden and others like it are not perfect, of course. The lack of emotions and situational awareness makes it difficult for the likes of ChatGPT and other programs to participate without an explicit prompt.

While there certainly are risks, they are, for the most part, relatively easy to manage. Let’s review the concerns cited most frequently in our focus groups:  Information leaks.  Sample bias.  And anchoring in the past.  

For the next decade or so, until a high level of digital literacy becomes a basic skill that virtually every professional possesses, directors will need to be trained in using AI.  How can we do it:  Create engagement.  Practice collective experimentation.  And maintain momentum.

show less