Wikit commits to more responsible Artificial Intelligence

2023 was the year of the Artificial Intelligence (AI) explosion. To understand the scale of this phenomenon, just take a look at ChatGPT and its 200 million unique visitors per month! However, ChatGPT is far from being the only tool to exploit Artificial Intelligence. There are many other conversational agents, such as Google's Gemini, and image-generating artificial intelligences, such as DALL-E or MidJourney, offering infinite possibilities. However, like all digital technologies, artificial intelligence is not just about immateriality, and the environmental challenges that accompany it are often overlooked.

• How big is the environmental and social impact of AI?
• What can AI companies do to reduce this impact?

In this article, we'll explore the impact of Artificial Intelligence on the environment, then look at the concept of Responsible Digital, before highlighting Wikit's commitment to more responsible AI.

The impact of Artificial Intelligence on the environment

The field of Artificial Intelligence is, by its very nature, energy-intensive. Indeed, the rise of Artificial Intelligence is accompanied by a growing demand for computing power, the environmental impact of which is considerable. Indeed, according to researchers at the University of Massachusetts, the training of a single generative AI model can release almost 300 tonnes of CO2 into the atmosphere. Alongside the issue of training AI models, data center infrastructures are also energy-hungry. These data centers, needed to store the data feeding AI algorithms, account for a significant proportion of global energy consumption. What's more, the manufacture and maintenance of the IT equipment required for AI also contributes to the technology's carbon footprint. So, while Artificial Intelligence offers unprecedented opportunities for efficiency and innovation, it also poses major challenges in terms of environmental sustainability.

Against this backdrop, more and more companies developing technologies using Artificial Intelligence are embarking on optimization initiatives to reduce the environmental impact of Artificial Intelligence as much as possible. Many organizations, for example, have made huge strides in training large language models on smaller databases, or are using models with fewer parameters.

Responsible digital technology: a path towards more sustainable Artificial Intelligence

Si la transformation digitale portée par l’Intelligence Artificielle révolutionne les entreprises, le numérique contribue malheureusement à hauteur de 3 à 4 % aux émissions mondiales de gaz à effet de serre et génère 2,5 % de l’empreinte carbone nationale à l’heure actuelle. Ces chiffres s’accompagnent de l’accroissement de la consommation d’énergie et des émissions globales de gaz à effet de serre, mais également d’autres problèmes tels que la pollution et l’épuisement des ressources. Face à ces enjeux, le concept de Numérique Responsable s’impose comme une démarche d’amélioration continue visant à réduire l’impact écologique, économique et social des technologies de l’information et de la communication, notamment de l’IA.

The Responsible Digital Labellabel, supported by the initiative of the same name, aims to encourage companies to adopt more responsible practices in the design, production and use of digital technologies. By encouraging eco-design, reducing energy consumption and promoting digital inclusion, this label helps to shape a digital world that is more respectful of the environment and human rights. It therefore addressesthe 3 pillars of sustainable development: people, planet, profits.

The Responsible Digital Label guarantees the credibility of a responsible digital approach, and the labeling process is therefore demanding. To obtain the label, an audit is carried out by an external auditor, and the label is awarded by a labeling committee of Green IT experts. This guarantees the credibility of an organization's commitment. Based on the results of the assessment, the organization defines a Responsible Digital Action Plan, which it undertakes to implement over a 2-year period. If the organization obtains more than 500 points out of a possible 1000, an independent labeling committee examines the action plan and decides whether or not to award the label.

Wikit, committed to more sustainable Artificial Intelligence, awarded the Responsible Digital Label

As a specialist in the creation of chatbots with Artificial Intelligence, we therefore wanted to obtain the Responsible Digital Label, to affirm our commitment to a more sustainable use of Artificial Intelligence. By integrating sustainability and ethical criteria into the development of our product, we are taking a socially and environmentally responsible approach, while offering our customers an innovative, high-performance solution.

Après un audit début 2023 pour réaliser un « état des lieux » de l’impact IT de l’entreprise et de notre solution, nous avons pu déterminer une feuille de route Green IT et ainsi commencer à réduire l’impact environnemental de l’IT au sein de notre organisation. Par la suite, en suivant pas-à-pas le référentiel Numérique Responsable, Wikit s’est engagé à mettre en œuvre des stratégies qui permettent d’adopter une démarche de conception responsable de sa solution d’IA générative, de faire évoluer les politiques d’achats ou encore d’offrir une seconde vie aux équipements.

Among other things, we have made the following commitments:

• Choice of hosting provider OVHcloud, committed to reducing its carbon footprint through the French location of its datacenters;
• Accessibility of our solution according to the Référentiel général d'amélioration de l'accessibilité (RGAA ), which aims to make digital content and services understandable and usable by people with disabilities;
• Prioritize team members' travel by train, rather than by car or plane;
• Donation of old IT equipment to associations working to promote re-use.

Committed to continuous improvement, we intend to pursue our efforts to take concrete steps to reduce our environmental and social impact.

Conclusion

In conclusion,Artificial Intelligence represents both a formidable lever for innovation and a major challenge in terms of environmental sustainability. However, thanks to initiatives such as the Responsible Digital Label and the commitment of many companies to minimize the impact of new digital technologies, it is possible to reconcile technological progress with social and environmental responsibility. In this context, Wikit is a committed player in the promotion of a more responsible Artificial Intelligence, at the service of a more sustainable and equitable future.