Between risks and opportunities, what place for sustainable AI?

June, 2024

Artificial intelligence (AI) is opening up a new paradigm, a form of new dependency. The flow of new uses is having a major impact. In this context, a sustainable use of AI remains conceivable, but not without reshaping our relationship with digital technologies.

“We feel a fragility, a great weakness on the one hand, and a great excitement on the other about everything that seems possible to do with AI”. Vincent Courboulay, associate professor at La Rochelle University and co-founder of the Institut du Numérique Responsable in France, explains this great ambivalence in us about AI. “One morning, if we get up on the right foot, we say to ourselves: we're going to remake the world! Another morning, if we get up on the wrong foot, we say to ourselves: the world is going to remake us!” he adds.

In his view, AI is a key component in the challenges of responsible digital technology, because “AI is a concentrate of digital impacts”. He continues: “Hundreds of thousands of people in precarious situations are labelling AI, hundreds of thousands of people are working to create the electronic chips needed for AI, hundreds of millions of people are affected in their jobs, and hundreds of millions of equipment are entering ever shorter life cycles in order to have ever more AI on hand...”.

This balance sheet also includes the environmental impact of the electricity and water consumption of the data centers needed to run AI. Despite this environmental and social footprint, companies see many opportunities for development.

AI in the enterprise: opportunities

In Vincent Courboulay's view, these opportunities lie above all in what are known as datalakes. With datalakes, generative AIs work on small, company-specific data fields. This greatly increases the precision of answers and their relevance to the company. 

“I don't believe in the notion of LLMs [Large Language Models], which would position AI as a global super-intelligence capable of responding to anything and everything”, he explains, “I think it's more interesting to believe in smaller, specialized, expert AIs, designed specifically for the company, and by the company”. In his view, this is a good way of making data speak for itself.

SustAIn.brussels, a European digital innovation hub in Brussels which regroups several actors like two universities in Brussels, ULB and VUB, and also Agoria, the Belgian equivalent of Numeum, helps companies take into account the footprint of AI and guides them towards more sustainable use. Evaluating the data used is key. 

“At SustAIn, we know that AI is an important technology that impacts a lot of our daily life processes” explains Ludwig de Locht, project manager at sustAIn.brussels, “we are convinced that we need to put technology to work in a responsible way, to make it sustainable not only economically, but also environmentally and socially.”

“Artificial Intelligence is just a toolbox” he continues, “the first step is to determine the adoption value of the change and see what technology could be useful. AI can improve many processes and be very useful for companies”. For Ludwig de Locht, the important thing is to look at the overall effort it takes to achieve a result, beyond the direct environmental impact of AI: “How many Google searches are needed to complete a goal? If in a single prompt on ChatGPT I can do it, the question of reducing the impact in this specific case is worth asking.”

Towards sustainable AI

In the eyes of Vincent Courboulay, a member of the GreenTech Forum Brussels 2024 program committee, a collective effort is required to make data speak for itself within a responsible framework. “Everyone has to take on a small part of the responsibility,” he says, “Responsible AI will exist if we work as a collective: elected representatives, local authorities, citizens, NGOs, businesses. We need to organise the means to get everyone around the table.” For him, it's also a way of creating a collective choice around the destination of public money: “which startups does public money finance?” he asks.

For Vincent Courboulay, the notion of understanding the world around us is also a key factor in moving towards sustainable AI. Reconnecting with nature. “We're putting ourselves outside the world! We talk about the environment as if we were outside it,” he argues, ”we don't see what surrounds us as something to which we belong. 

This is one of the reasons why he believes generative AI has a footprint that goes far beyond the impact of a prompt. “To save an hour or two, we have an impact on someone far away from us”, he continues, “in a society where the notion of effort is disappearing, that's problematic. Maybe if we find the means to get out of the comfort society we're in, we'll return to a more supportive society tomorrow”.

Beyond society's relationship with generative AI, it's also the relationship that organisations have with AI in particular and digital in general. Ludwig de Locht explains this challenge: “At SustAIn, we are convinced that responsible digital transformation must be an integral part of any company, whether it's very small or very large”. 

For Ludwig de Locht, companies need to strike a balance between economic, social and environmental issues to be sustainable. “You can't be 100% sustainable in the beginning, but you can improve over time and become more mature. The important thing is balance and progression,” continues the man who, like Vincent Courboulay, is also a member of the GreenTech Forum Brussels 2024 program committee, “that's the vision we try to bring to companies”.

So it's also beyond the simple question of AI's footprint that the question needs to be raised. If companies, like society as a whole, do not find the means to enter into a real ecological transition, then it will be difficult to use AI responsibly, and it will be difficult to create sustainable AI-based digital services. 

The question of responsible digital transformation of companies therefore largely encompasses the challenges of using AI, which is still, for the time being, only one tool among many. As Vincent Courboulay concludes: “Let's not forget that AI is inoperable without human intervention! Autopilot can't work on its own. So it's up to us to decide what kind of AI we want tomorrow!


Ludwig de Locht is Project manager at sustAIn.brussels and Digital Transformation Ambassador at Sirris. He leads the organisation of GreenTech Forum Brussels for sustAIn.brussels and is part of the event program committee.

Vincent Courboulay is an associate professor at La Rochelle University, co-founder of the Institut du Numérique Responsable (INR) in France and is part of the GreenTech Forum Brussels 2024 program committee.

Article written by Rémy Marrone for GreenTech Forum Brussels

GreenTech Forum Brussels is the new Tech and Sustainability event.
Co-organised by the Belgian Institute for Sustainable IT, Numeum and SustAIn.brussels, it will take place the 18 et 19 of June 2024 at La Maison de la Poste in Brussels.
More information