rethink sustainability

    How can AI speed up the decarbonisation of business?

    Article published on Le déCLIC® responsable in partnership with Le Figaro on 9 October 2023

    Algorithms are enabling companies to speed up their energy transition and comply with regulations by processing and optimising their data.

    Since July 2022, businesses in France with more than 500 employees have had to prepare a greenhouse gas emissions assessment (BEGES) every four years. Those with more than 250 employees will need to do so from 2025, and eventually even small and medium-sized enterprises will have to meet this requirement. These carbon footprint calculations are largely based on information the companies already have available, such as invoices, but they will soon have to start going further by asking questions of their suppliers, customers and employees. The volume of data to be processed is increasing, and some of it is not digitised. This is creating a serious challenge for businesses, but artificial intelligence (AI) could simplify and speed up this process, and even support companies in their long-term decarbonisation strategies.

    Read also: AI in financial services: is this the new dotcom bubble?

    But before this can happen, an information gathering phase is needed – an essential step for training AI machine learning algorithms. “This is the first kilometre of a marathon, but the course still needs to be mapped out,” says Gwendal Bihan, co-founder and CEO of Axionable, a management and technology consulting firm established in 2016. “For AI to be useful, a great deal of data gathering work needs to be undertaken beforehand. Only larger groups with a few years of carbon footprint data at their disposal are able to take informed action.” And yet the clock is ticking.

    The Paris Agreement, adopted by France and the European Union in 2016, aims to keep the rise in global temperatures below 2 degrees Celsius, and ideally not above 1.5 degrees Celsius. It requires radical changes – a 40% reduction in carbon footprints by 2030 compared to 1990, to achieve a carbon-neutral world by 2050 (a goal set by the national low-carbon strategy of the French Ministry of Ecological Transition). In other words, tomorrow, on a company scale.

    Watch the interview with Antoine Rostand, President and co-founder of Kayrros: 

    Predict, prevent, cure

    To run this marathon under the best possible conditions, businesses large and small are using specific, simple and increasingly accessible methodological tools. These systems, which cost hundreds or thousands of euros a year, automate carbon footprint measurements (Traace, Sweep, Carbometrix, Sami, Greenly, etc.) but without the use of AI.

    For Axionable, however, AI is the best way to chart a strategic course ahead: to predict, prevent and cure in order to anticipate future regulatory and climate changes. The consulting firm, which prides itself on an annual growth rate of 10%, combines analysis with data collection from large private and public organisations (industrial groups, nuclear and other energy companies, investment funds, insurance firms, the media, etc.).

    It is preparing these institutions for a changing world – rising temperatures, megafires, heatwaves and droughts. “Much of our work focusses on the 2030-50 horizon to help organisations adapt – change their industrial facilities, their operating methods, their suppliers and possibly their production sites,” explains Bihan. Added to this is a growing concern among investors and finance companies, who want to invest in greener companies and expect them to show a strong commitment to green issues.

    Today, millions of data points are available... so rich and plentiful that processing them exceeds the capacity of the human brain. But not that of AI

    Influence on public policy

    “AI makes it possible to do things that were impossible to do before,” says Jacques Sainte-Marie, a researcher at Sorbonne University and at the INRIA (National Institute for Research in Digital Science and Technology), of which he is Deputy Scientific Director, responsible for issues related to digital science and the environment. And he reminds us that “without digital data there would be no report from the IPCC [Intergovernmental Panel on Climate Change of the United Nations]”.

    Read also: AI’s infrastructure points to a first phase of monetising

    Today, millions of data points are available (satellite images, sensors measuring temperature, humidity, concentration of greenhouse gases in the atmosphere, etc.). This data is so rich and plentiful that processing it is now beyond the capacity of the human brain. But not that of AI. “Computers have been programmed to sort through this deluge of information and extract the relevant aspects,” explains Philippe Ciais, a researcher in the Climate and Environmental Sciences Laboratory at the French Alternative Energies and Atomic Energy Commission (CEA). This allows scenarios for the years and decades ahead to be refined, hypotheses to be tested, and decisions to be taken.

    Analysis of satellite images

    These climate models and projections provide reliable data that helps decision-makers responsible for implementing public decarbonisation policies, such as the recommendations of the IPCC. These projections are partly drawn from geospatial surveillance and detection technologies by private companies, including Kayrros. The French company is a global leader in its sector and was named one of the world’s 100 most influential businesses in 2023 by Time magazine.

    Thanks to Kayrros, leaks of methane – which has 80 to 100 times more global warming potential than CO2 – are detectable for the first time... from space

    Thanks to Kayrros, leaks of methane – which has 80 to 100 times more global warming potential than CO2 – are detectable for the first time... from space. Its algorithms analyse satellite images from the European Union’s Copernicus programme and from other publicly available information sources, such as businesses’ activity reports or information obtained from social media. They assess greenhouse gas emissions, the size of forests and trees, their CO2 storage capacity, the scale of deforested areas, and they track modes of transport. They also calculate energy expenses and measure economic activity; they have already enabled Australia to significantly reduce methane emissions from coal mines, and helped the United States fine site operators who are breaching regulations. A dedicated platform, Methane Watch, has also been established to monitor methane leaks. Kayrros has around 100 customers around the world, including sizeable private players and governmental and international organisations.

    Read also: Davos 2024: A window into the reorganisation of large-scale flows of capital around the transition

    The new generation of satellites that the European Union is investing in means it should be possible to obtain accuracy to the nearest metre in around the next two years. Outbreaks of fire could then be visible from the very first flames. Even the smallest tree will be counted and the records will be more complete. But to change the game “we still lack real political leadership, climate governance at a global level, genuine awareness of the issues, border adjustment systems, and we need to ban uncontrolled energy imports,” says Kayrros Founder and CEO Antoine Rostand. “Global CO2 and methane trajectories have not moved. What we are seeing is total collective inefficiency.” This means AI is not the miracle solution, according to Philippe Ciais. “It will help us find solutions to combat climate change, but it will not reduce emissions for us,” adds the researcher, who is also a member of the French Academy of Sciences. “Ultimately, it’s the politicians who will decide.”

    The potential of AI to tackle environmental challenges is huge, but only if the human behind it takes “the right moral and ethical decisions”

    The environmental bill of AI

    In France, the digital sector generates more greenhouse gases than the aviation sector: 2.5% compared to 1.5% of the total. Its carbon footprint could treble by 2050 if the current trajectory continues.

    AI needs extremely powerful processors, supercomputers and servers to train algorithms and store data, and this power comes from electricity that may derive from fossil fuels. This has led to the concept of “frugal” AI and resilient digital technology that “take into account the finitude of the world”, says Jacques Sainte-Marie. “We need to develop environmentally responsible digital technology, with software that’s smaller, less power-hungry, with less obsolescence, with slow tech, microprocessors that aren’t always state-of-the-art, preliminary data selection, etc., because this is also an economic lever. In a few years’ time, businesses doing this will be the big winners; they will be the most competitive, they will be sovereign.”

    The potential of AI to tackle the environmental challenges is huge but only if the human behind it takes “the right moral and ethical decisions”.

    Important information

    This document is issued by Bank Lombard Odier & Co Ltd or an entity of the Group (hereinafter “Lombard Odier”). It is not intended for distribution, publication, or use in any jurisdiction where such distribution, publication, or use would be unlawful, nor is it aimed at any person or entity to whom it would be unlawful to address such a document. This document was not prepared by the Financial Research Department of Lombard Odier.

    Read more.


    let's talk.