Dealing with opportunities and risks of AI in the boardroom

In this capacity she also acquired practical expertise in relation to impact of artificial intelligence and how to deal with it in the boardrooms of profit and non-profit companies. She was also a panelist on the GUBERNA National Member Forum that sought answers to how a director can expertly navigate innovation, risk and responsibilities with AI applications. These applications are currently experiencing a boost in various enterprises and organisations. We are grateful that she agreed to share her useful experiences and advice for directors in various companies in this GUBERNA Directors Sparkles. 

"Data-driven enterprises and AI technologies are 'enablers' of the digital transformation in our society." – Katrin Geyskens 

Over the past year, so much has been said about Chat GPT, Bing, Dall-e and other AI applications. For a better understanding, can you briefly outline what Artificial Intelligence is and what AI applications actually do?  

Artificial Intelligence is software that uses data science (such as statistics, 'machine learning', 'deep learning', etc.) to mimic human intelligence in order to produce output, including content creation, prediction, recommendation generation or even decision making.   

These AI applications are everywhere and can be roughly divided into three categories: 

• Simple applications that incorporate AI (which we are not always aware of): e.g. spam filter on email, 'lane assist' in the car, optimised ordering from bol.com and having the parcel delivered, etc. 

• Highly specialised applications for which 'cutting-edge' AI algorithms are specifically developed, which support users in a simple way: for example, brain scan analysis to support radiologists who get a 'simple' report on observations they cannot see with the naked eye (Icometrix) 

• Generative AI intended for the 'consumer'/wider society: e.g. Chat GPT (97% of companies currently use it according to a Mc Kinsey survey, with only 20% of these companies having an appropriate policy on this use). 

Op welke manieren kunnen ondernemingen en hun bestuurders te maken krijgen met AI-applicaties en het gebruik van AI? 

First of all, many enterprises use AI and AI applications. They use it in a variety of ways and to different degrees. 

Use in horizontal processes across all businesses can be characterised as a natural evolution of software. Who doesn't use 'finance forecasting', 'social media tools' for marketing purposes, chatbot', etc.? The level of risk is low and these applications are widely distributed.

Secondly, of course, there are companies that develop and commercialise AI applications based on their own algorithms and with their own training data. This is the case mostly in innovative and often 'venture-backed' companies such as Ghent-based FEops, which makes a 'digital twin' of the patient's heart. This software is then used to prepare the placement of a new heart valve in a specific patient. 

How can AI be used in the boardroom?  

Most boards already unknowingly use AI in their preparation, for example because it is used by the finance department to do forecasting ('assistive AI').  

However, some applications exist that go even further and support the Governing Body in their preparation. One example is Ghent-based company Trensition, which uses AI to deliver trend analysis specific to a particular company, sector or topic. This can feed boards with strong or weak signals of trends to either or not act on, to adjust strategies and respond to or not earlier than the competition.  

The question is raised whether generative AI such as ChatGPT can also help individual directors prepare for the meeting. You might ask why not make a certain investment with the aim of bringing out data as enrichment and a richer debate when a decision needs to be made. If you ask these questions and use generative AI, it is absolutely recommended to use the paid and professional version, if not, the data and input can later become public . It might be a good idea to try out, as long as it is restricted to enriching the debate and not being used as an absolute truth with the sole intention of stopping an investment. So smart queries to AI applications and the input generated from them might allow a director to easily gain more insight.  

In academia, there is also talk of using full 'autonomous AI' or a so-called 'robo-director' or robot director: an AI application that can make decisions like an autonomous director, after being trained to do so. However, this does not seem to be for immediate use yet. How to train a 'robo director', safeguard them from 'bias', make them function collegially in the diversity of the boardroom, etc.?  

What does this evolution mean for directors' responsibilities?  

Depending on the category of AI applications and the degree of their use/development, the director's responsibility will differ. A higher-risk category (e.g. applications used in the medical field, or applications used in the recruitment process) and self-developing AI applications based on training data requires a more intense supervision and an adapted 'framework' for compliance, risk management and audit.  

This is not a one-off thing, it requires continuous questioning and testing of the adequacy of the 'frameworks' in function of the evolution inside and outside the company.  

It is too early to define universally applicable rules now, but a good frame of reference is the EU AI Act, which is likely to come into force in 2026 .  

Should the composition of the board be adjusted in the context of by these technological innovations? Should the chairman be an AI expert or is it appropriate to recruit a director with AI expertise? 

There are not many true AI experts today and, moreover, a director is not supposed to be an expert. However, there should be sufficient basic knowledge and hopefully someone in the organisation who can go deeper and look into the algorithms and 'sourcing' of data, this is not the role of the director. AI deserves a particular focus of supervision for updating policy frameworks of compliance, process & risk management and audit and, if necessary, monitoring adequate KPIs. 

Do you have any suggestions for directors who want to learn about Artificial Intelligence for well-informed directors? 

I can recommend the following books: ‘Mens versus machine. Artificiële Intelligentie ontrafeld’ by Geertrui Mieke De Ketelaere and ‘Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence' by Kate Crawford. There are also useful websites such as e.g. Flanders AI Research Program and Artificial Intelligence Digital Future. The latter is very useful for SMEs. 

Exchange with fellow directors is of course also very instructive, for example during National Member Forum or other GUBERNA initiatives.