France/ AI at the service of sustainable development: a recognized potential but little exploited by companies
Artificial intelligence (AI) is sometimes presented as the miracle solution to the challenges of sustainable development. Industry professionals are not so sure. Our recent survey of over 400 IT managers in France and the UK, the results of which were presented to the French Senate, reveals a contrasting landscape.
The "AI and Business" barometer, which emerged from this survey, certainly shows that businesses are optimistic overall, with 54% of respondents believing that AI has a positive impact in terms of sustainable development. The most cited areas of application are the reduction of greenhouse gas emissions, waste management and supply chain optimization.
However, perceptions vary widely from one area to another. Among the UN's Sustainable Development Goals (SDGs), health, education, energy and industry are perceived as the sectors where AI could have the most significant impact. In healthcare, it is seen as a promising tool for improving diagnosis and personalizing treatment. In education, it could enable learning that is more tailored to individual needs. In energy, the technology is seen as a means of optimizing the production and distribution of renewable energies. In industry, it is seen as a lever for developing more sustainable processes.
However, the "AI and Business" barometer also highlights blind spots. Certain SDGs, such as the preservation of aquatic and terrestrial life, are rarely cited as areas of application. Could a more global approach be possible?
Companies insufficiently aware
In fact, only 36% of companies track the energy consumption of their AI systems. Even more worryingly, just 29% measure their net AI-related greenhouse gas emissions. This discrepancy between stated ambitions and the reality of practices raises many concerns. It often translates into disappointing environmental performance. Microsoft is a case in point. While the company committed in 2020 to halve itsCO2 emissions by 2030, its Sustainability Report 2024 sustainability report reveals that its indirect greenhouse gas emissions have increased by 30.9% compared to 2020. This increase is mainly due to the expansion of data centers required to support AI technologies.
Companies also face significant ethical challenges. Only 28% have tools to detect or resolve AI-related ethical issues. 18% have already had to stop or adjust an AI project for ethical reasons.
The main concerns are data confidentiality, transparency of model decisions and social impact. These issues are particularly important when it comes to using algorithms for sustainable development objectives, which often involve sensitive data and decisions with a significant impact on populations. This is the case, for example, of using AI to optimize energy consumption in smart cities.
The study also highlights a glaring lack of training, not only on technical aspects, but also on the ethical, social and environmental implications of AI. Currently, only 30% of companies offer training on its ethical use. This gap can have far-reaching consequences. Without a good understanding of the ethical and environmental implications of AI, companies risk developing solutions that, while innovative, could have unexpected negative effects on society or the environment.
From awareness to action
For AI to live up to its promise in terms of sustainable development, several challenges identified in the barometer need to be addressed. First and foremost, we need to be able to accurately measure their environmental impact, and develop standardized methods for assessing energy consumption and greenhouse gas emissions throughout the lifecycle of AI systems. Microsoft's example shows the importance of this precise measurement for achieving climate objectives.
Secondly, efforts must be made to systematically integrate ethical considerations. This means setting up formal frameworks and developing tools to detect and correct biases in AI models. We also mentioned the need for massive employee training, which should be interdisciplinary, combining technical skills in AI with an advanced understanding of sustainable development issues.
Finally, we need to develop close collaboration with political decision-makers to establish frameworks that promote responsible AI. This could, for example, include incentives for companies that develop sustainable AI solutions.
The study shows that we are at a turning point. Companies recognize the potential of AI for sustainable development, but they now need to move from awareness to concrete action. The development of sustainable AI can also be achieved through research. A recent study demonstrated the superiority of AI models over traditional models in predicting greenhouse gas emissions by states.
Source: theconversation.com/
