Artificial intelligence and the environmental component of corporate social responsibility

Sarah Beret

p. 129-146


Texte intégral

1Can artificial intelligence (AI) contribute to the environmental component of corporate social responsibility (CSR)? Can AI push companies to be more responsible? The notion of AI can be defined as a body of theories and techniques executed for the purpose of designing machines that are able to simulate human intelligence1. In this paper, we will explore how these machines may lead companies to be more environmentally responsible.

2To understand the topic, it is essential to define the concept of CSR. This concept has its roots in the beginning of the twentieth century with the school of thought “Business and Society”. This movement strove for a vision of a company that addressed common concerns within the society2. In 1953, Howard Bowen refined this idea, developing an ethical management approach3. Subsequently, John Elkington4 advocated for the joint inclusion of economy, environment, and social rights. These elements form the basis of CSR.

3This term was defined for the first time in 2001 in the green paper of the European Commission, “Promoting a European Framework for Corporate Social Responsibility5. Here, CSR is defined as a “concept whereby companies integrate social and environmental concerns into their business operations and in their interactions with their stakeholders”. In light of this definition, CSR is the emanation of the will of the company, that sets social and/or environmental objectives resulting in the adoption of behaviors that exceed the expectations of society. Ten years later, the European Commission broadened this definition. Currently, CSR is defined as “the responsibility of enterprises for their impact on society6. This expansion notwithstanding, social and environmental concerns remain the backbone of CSR. Our interest will focus on its environmental component. In other words, a company should master the environmental impacts of their activities. At first glance, this consideration isn’t the main objective of a company whose primary function is to maximize profit for its shareholders. Nevertheless, the two objectives aren’t mutually exclusive. Indeed, CSR assigns an image of corporate citizenship, improving the reputation of a company often resulting in the investors perceiving a more viable company, and clients being attracted by the company’s ethics. As such, these factors ensure better economic outcomes for companies that implement components of CSR.

4To foster better economic outcomes, the execution of CSR can be enhanced by AI. It can provide new pathways for companies to achieve their environmental objectives. However, the current state of AI for the environmental component of CSR needs to be improved in order to provide more and better solutions to companies. The goal of this article is to highlight the extraordinary opportunity that AI represents for the environmental component of CSR (I) and to suggest improvements that can maximize AI’s potential (II).

I. Artificial Intelligence: an extraordinary opportunity for the environmental component of CSR

5To understand the opportunity offered by AI, the contents of the environmental component of CSR should be developed (A). This should be juxtaposed with the practice of AI (B) and to the legal regulation of AI (C).

A.The content of the environmental component of CSR

6It is crucial to point out that the content of the environmental component of CSR is limited compared to the social component. However, several international texts cover the environmental pillar. The first text is the Global Compact created by the United Nations7 in 2000. Ten years later, the Organization for Economic Co-operation and Development (OECD) adopted Guidelines for Multinational Enterprises8. The same year, the International Organization for Standardization (ISO) implemented the ISO 26000 standard9. Finally, United Nations members passed Sustainable Development Goals in 201510. This body of international texts illustrates the diversity of organizations that have tried to provide their vision of the environmental dimensions of CSR. As far as their content is concerned, some texts such as the Global Compact develop several general principles. It highlights an objective but doesn’t offer companies solutions for reaching their proposed goals. In contrast, texts such as the Guidelines for Multinational Enterprises and the ISO 26 000 standards attempt to propose more practical solutions. Lastly, the Sustainable Development Goals are more practical.

7A comparative reading of theses texts demonstrates some general principles such as, “Business should undertake initiatives to promote greater environmental responsibility”11 and business should “contribute to (…) environmental (…) progress with a view to achieving sustainable development”12. More specifically, companies should “assess, and address in decision-making, the foreseeable environmental (…) impacts”13 and “continually seek to improve corporate environmental performance”14. To be more practical, the environmental pillar of CSR could be comprised of several elements: management of resources, anticipation of environmental risks, and production of green technologies15.

8Concerning the management of resources, several resources should be considered: water, waste, and energy. Regarding water management, companies are supposed to “ensure the sustainable management of water16. For waste management, enterprises should practice recycling of paper, plastic, glass, and aluminum17. As for energy usage, businesses should optimize their energy consumption by, for example, using low consumption equipment and bulbs18 and maximizing their use of renewable energy.

9When anticipating environmental risks, companies should execute a system of environmental management19. This should include “collection and evaluation of adequate and timely information regarding the environmental, health, and safety impacts of their activities”20 and “maintain contingency plans for preventing, mitigating, and controlling serious environmental damage from their operations, including accidents and emergencies”21.

10With the environmental component of CSR delineated, it should be juxtaposed with currently used AI systems to determine if they can provide real solutions for companies in support of their environmental objectives.

B. AI in practice: a real solution for the environmental component of CSR

11Several AI innovations already help enterprises to be more environmentally responsible22. Specifically, the area that shows the highest potential for AI contribution is the management of resources, and especially the optimization of energy.

12Concerning water management, IBM23 and VEOLIA24 offer customers a monitoring platform which ensures the integration and optimization of data linked to water management. The company EMAGIN25 has developed AI which can analyze and learn from data collected by water sources using sensors. This system can predict future water consumption and make recommendations to maximize efficiency, reducing waste of water resources.

13Several companies provide AI solutions for waste management. For example, IMMERSIVE ROBOTICS26 created the robot Baryl. Through progressive learning, the robot moves toward a customer when it detects that the customer wishes to dispose of a piece of trash. When the robot is full, it returns to its base where it alerts a human that it needs to be emptied. In another example, BULKHANDLING SYSTEM27 using vision systems and deep learning algorithms, produced a robot that sorts by moving its robotic arm toward a required element. These are just two examples of AI solutions that help companies improve their waste management.

14AI solutions are already in use for the improvement of energy consumption. The company TWAICE28 has developed predictive analysis software that works based on a digital twin optimizing the use of lithium-ion batteries. Furthermore, ENERGIENCY29 implemented an AI platform that analyzes energy performance. As a result, customers can save up to 20% on their energy bill. The company SENSING VISION30 created an enhanced AI model to identify, analyze, and correct energy anomalies in their buildings. It is based on a network of low consumption sensors and AI algorithms that process collected data. The company guarantees that its clients will save between 20% and 40% on their energy bills. The AI solutions developed by TWAICE, ENERGIENCY and SENSING VISION support energy optimization and, in turn, contribute to reduced carbon emissions.

15AI also provides solutions for renewable energy. The company ELUM ENERGY31 established an integrated management and storage system for electric and solar energy. Their system is composed of solar panels, batteries, and a smart electrical module to reduce peak energy consumption. In the same sector, the company SAUREA32 created the first autonomous solar motor that directly converts solar energy into mechanical energy without electrical conversion. This process boosts the reliability of solar installations in off-grid sites. These products may even encourage companies to use renewable energy.

16Furthermore, AI provides opportunities beyond what is planned stricto sensu by the international texts. For example, AI can help to improve a company’s air quality. To this end, PARTNERING ROBOTICS33 developed a robot named “Diya one” that fosters well-being at work. It analyzes the following indoor environmental quality parameters: air, energy, temperature, and luminosity by moving through every office, collecting fine particles from the air.

17AI can also help companies to be more environmentally responsible while improving their economic activities. An excellent example is in the farm industry. ORBITAL INSIGHT34 implemented a predictive data service based on analyses of images and data collected by satellites and drones. This service enables farmers to better manage their resources by providing information about the best time to harvest their crop. In addition, this service helps farmers to determine the retail sales forecast. More broadly, AI is a solution that improves weather forecasts and helps to anticipate extreme weather events.

18AI is the confluence of an entire family of technologies that can provide a wide range of environmental and economic benefits. More specifically, AI enhances predictions, optimizes the allocation of resources, and personalizes the delivery of services35. These benefits are driving further research into the use of AI, providing opportunities for companies to be more environmentally responsible. Examples in France include: MAIA in Grenoble, PRAIRIE in Paris and ANITI in Toulouse36. Another example of the implementation of AI comes from the French Agency of Biodiversity, which has implemented an AI solution to better control environmental police37.

19In conclusion, these examples demonstrate the practical opportunities that AI represents for the environmental component of CSR. In fact, in the last several years, AI regulations have taken environmental protection into consideration.

C. The handling of the environmental protection by AI regulations

20The opportunities offered by AI for the environmental component of CSR are directly addressed in “the framework of ethical aspect of AI, robotics and related technologies38. More specifically, recital 55 of this regulation supports the establishment of non-binding guidelines to implement a method to assess the achievement of “the objectives of social responsibility, (…) environmental protection and sustainability”. This regulation is direct evidence of the opportunity that AI solutions offer for the environmental component of CSR.

21Other AI regulations don’t directly address the environmental component of CSR; however they take into consideration the subjects of AI and environmental protection. Several texts such as “The Ethics guidelines for trustworthy AI39, “The white paper on AI40 and “the regulation of the European parliament and of the council laying down harmonized rules on AI41 address concerns for these subjects, pointing out the opportunity AI provides for protecting the environment. These texts highlight that AI “offers possibilities for energy transition42 and “represent a great opportunity to support the mitigation of pressing challenges facing society such as (…) environmental pollution43 because it could “reduce humans’ impact on the environment and enable the efficient and effective use of energy and natural resources44. These European texts suggest that AI should be used to protect the environment45 and support the achievement of environmental objectives46.

22To contribute to environmental protection, the various European texts proposed several strategies including the promotion of research into AI solutions to global concerns. These include development of sustainability objectives that support environmental protection47. In addition, AI should promote a critical examination of resource management and energy consumption48. Finally, AI systems should be designed to make decisions that have a positive effect on the environment.

23Nevertheless, theses recitals are merely general principles lacking in actual scope. Currently, only two articles of the European Regulations address the environment. The first, Article 11, addresses the framework of the ethical aspects of AI. It focuses on the assessment of environmental durability. Its relevance is that it supports the evaluation of the environmental impact of an AI system by the national supervisory authority and other national or European sectorial supervisory bodies. For this purpose, these authorities have the power to take measures to mitigate and remedy the general impact of AI, “as regards natural resources, energy consumption, waste production, the carbon footprint, climate change emergency and environmental degradation49. Therefore, authorities charged with these evaluations have the power to propose improvements to the AI systems, promoting heightened environmental responsibility, leading AI developers to provide companies with more environmentally responsible AI and helping them to achieve their objectives of CSR.

24This assessment; however, conceals an obscure point concerning the scope of its application. Effectively, the evaluation is limited to high-risk AI. This notion is defined by the ethical framework of AI, “as significant risk entailed by the development, deployment and use of artificial intelligence (…) to cause injury or harm to individuals or society in breach of fundamental rights and safety rules as laid down in Union law50. This risk is assessed in the context of AI’s specific use or purpose, the sector in which it is developed, deployed, or used, and the severity of injury or harm that can be expected to occur51. These elements demonstrate that the notion of high-risk AI is eminently subjective, potentially resulting in different interpretations by each state in the European Union. Therefore, an AI system could be considered high risk in some states but not in others, resulting in AI systems being excluded wrongly from the evaluation. To limit these differences of interpretation, it is preferable that the European Union establishes guidelines employing a precise interpretation of high-risk AI needing to be assessed concerning their environmental durability. Thanks to these guidelines, more high-risk AI can be improved through assessment of environmental durability and companies using them could more easily achieve their CSR objectives.

25Several months after the development of the ethical framework of AI, the European parliament and the council proposed to unify their rules on AI. Specifically, Article 69 of the regulation52 addresses the discretionary nature of a company’s code of conduct. According to this article, “The Commission and the Board shall encourage and facilitate the drawing up of codes of conduct intended to foster the voluntary application to AI systems of requirements related for example to environmental sustainability”. Once more, the European Union remains evasive and doesn’t guide AI producing companies to take into consideration environmental sustainability in their strategies. Furthermore, the codes of conduct are discretionary: the companies may choose to adopt them or not. With such limited guidance, companies specializing in AI have no motivation to adopt a code of conduct and/or include environmental sustainability objectives in their strategy. To foster the adoption of codes of conduct in the AI sector, the European Union should provide an exemplar code of conduct with precise environmental sustainability objectives. This would make it easier for companies that produce AI to adopt a code of conduct, and to create more environmentally responsible AI systems that other companies can use to achieve their CSR objectives.

26These ethical and regulatory issues demonstrate the incredible opportunities that AI affords the environmental pillar of CSR. However, despite these incredible opportunities, there are still improvements that must be made.

II. Artificial Intelligence: areas for improvment to help companies achieve their CSR environmental objectives

27AI is poised to help more companies achieve their CSR environmental objectives. However, in order to do so, AI must address two issues: promotion of environmental data sharing (A) and the reduction of digital pollution caused by AI (B).

A. Promotion of environmental data sharing

28The paucity of environmental data is highlighted in reports conducted by the French Ministry of Economy53 and the French Deputy Cedric Villani54. These reports conclude that, except for satellite imagery and meteorological data, environmental data suffers from limited availability55. The collection of additional environmental data will help to develop more robust AI systems that in turn, will assist companies in achieving their CSR objectives. Therefore, it is crucial to establish legislation that encourages the collection and sharing of environmental data.

29This type of legislation is essential in view of the potential impact AI can have on the environment. For example, data generated from French electric meters (Linky) could help to optimize the consumption of electricity by providing better usage forecasts. Data analysis and decision-making on air pollution could be enhanced using AI resulting in more effective warnings about air pollution56. Moreover, the Open SolarMap57 project that studied cadastral, satellite, and contributive data, concluded that the value of AI in data collection and sharing could be immense.

30Access to environmental information in the European Union started with the Directive n°2003/04/EC58. In Article 2 of the regulation, “environmental information” was defined broadly to include, “the states of the element of the environment” and “factors, such as substances, energy, noise, radiation or waste, including radioactive waste, emissions, discharges and other releases into the environment, affecting or likely to affect the elements of the environment”. Article 3 goes on to state that member states should ensure that environmental information be made available to the public59. This provided a good opportunity to ensure release of environmental data; however, this old regulation fails to view environmental data through the lens of AI. By introducing many exceptions including the confidentiality of commercial or industrial information and the confidentiality of personal data and/or files relating to a natural person, Article 4 excludes disclosure of a lot of information under the guise of confidentiality. De facto, the regulation seems to complicate the collection and sharing of precise environmental information that could potentially help companies develop AI solutions for the protection of the environment. Rather than enhancing the public’s access to environmental data, the aim of the regulation was to promote informing the public of environmental dangers.

31For the last 5 years, the subject of data protection has received significant attention by the European Union. The most well-known regulation is 2016/67960. It concerns the “protection of natural persons with regard to the processing of personal data and on the free movement of such data61. Although it regulates the protection of personal data, it does not address the subject of environmental data. Regulation 2018/1807, adopted 2-years later by the European Union, addresses the framework for the free flow of non-personal data62. It supports the free movement of data within the European Union with two main caveats: the prohibition of data location requirements63 and to encourage and facilitate the porting of data64.Unfortunately, this regulation doesn’t include specific language for the collection and distribution of environmental data.

32The first regulation which specifically dealt with environmental data was Directive 2019/1024. It focused on open data and the re-use of public sector information65. It created the notion of “high-value datasets66 that includes language where re-use is associated with “important benefits for the society, the environment67. Annex 1 of the directive cited six categories of high-value datasets, one of which is, Earth observation and environment, clearly meant for environmental protection. Although some exceptions were stipulated, these data-specific rules allow for free re-use of data68 and its availability in a readable format69. However, directive 2019/1024 only addresses data included in documents held by public sector bodies or from public undertakings70, excluding its application to private environmental data.

33The latest proposed regulation on European Data governance71 seems to expand the scope of Directive 2019/1024 in that it addresses data maintained in the public sector but protected by confidentiality. Nevertheless, it isn’t specific to environmental data, or the concept of high-value datasets treated previously. In its explanatory memorandum, it highlights the sector-specific legislation on data access stating that it “is in place and/or under preparation to address identified market failures in fields such as (…) environmental information72. Their explanation appears to already indicate that environmental data is threatened in directive 2003/04 EC73. Finally, these proposed regulations don’t appear to speak to the circulation of environmental data.

34Proposed regulations by the European parliament and the council addressing harmonized rules on AI seem to encourage data use contributing to environmental protection. Specifically, Article 54 of this proposed regulation states, “Further processing of personal data for developing certain AI systems in the public interest in the AI regulatory sandbox74. It authorizes the lawful processing of personal data collected for the purpose of developing and testing certain innovative AI systems. Specifically, this is allowed if the innovative AI systems are developed for the safeguarding of substantial public interest, “a high level of protection and improvement of the quality of the environment75. Irrefutably, this proposed regulation supports the use of personal data to develop AI systems contributing to environmental protection. Although the scope of this article encourages the use of personal data for the development of AI if these data could be of benefit to environmental protection, does data linked to a natural person contribute to the preservation of the environment? Only experience will give us the answer; however, one could conjecture that personal data won’t be the best source to create AI systems that protect the environment.

35These regulations addressing AI and/or data don’t address the communication of non-personal data owned by companies. These could be some of the most interesting sources for the development of more environmentally responsible AI systems. Some politicians support the opening of private data in a detailed way76. Specifically, the opening of private data owned by companies should be realized within the framework of execution of sectorial challenge. This could be an interesting first step in adopting real regulations concerning environmental data owned by private companies. This type of regulation should also address the communication of data in the specific framework and grant security and economic compensation to companies.

36Nevertheless, the circulation of environmental data isn’t the only problem facing the relationship between AI and the environmental component of CSR. Another issue is that AI can cause digital pollution.

B. Reduce the digital pollution caused by AI

37Although AI could provide benefit to the environment, it is possible that AI could harm it as well. Indeed, the relationship between AI and the environment is ambivalent77. AI can contribute to sustainable development, but it can also present some environmental risks, especially if it drives customers to increase consumption. Several entities like NGOs78, members of civil society79 and web actors80 bring attention to the risk of digital pollution caused by AI. Theoretically, digital pollution is divided into two parts81. The first is in regard to the physical materials linked to the environmental impact of operating the internet, its infrastructure, and the manufacturing of digital equipment. This involves a very high consumption of non-renewable resources and the creation of waste during the production process and their use. The second part of digital pollution is linked to the environmental impact concerning the operation of the internet network itself. This includes storage; sending email, use of search engines, use of the cloud, and streaming. The development of AI results in an increase in the volume of data stored and exchanged, an increase in computational volume, and pressure on companies to replace old equipment in order to improve performance. For instance, AI uses a graphic processing unit (GPU) that requires the use of silicon in the manufacture of electronic circuits and transistors. GPU’s also consume large amounts of energy. Thus, it is essential to reduce AIs potential adverse effects in order to maximize its benefit to the environment and so that companies can achieve their environmental objectives of CSR.

38In practice, several companies have tried to reduce the environmental impact of their data centers. For instance, Google employed AI technology to anticipate times during which their data centers were the busiest82. This operation helped to optimize the cooling of their data centers and led to a 40% reduction in energy consumption. Huawei uses a similar process to great effect. This supports the concept of legal regulation of the environmental effects of AI.

39Recent regulations concerning AI have tried to limit the environmental risk caused by AI. These regulations fall into three categories: analysis of the supply chain, product life cycle, and liability for environmental damages. The Villani`s report, “the framework of ethical aspect of AI, robotics and related technologies”, “The Ethics guidelines for trustworthy AI”, and “The white paper on AI” all agreed on the importance of the greening of the supply chain83. Moreover, “the framework of ethical aspect of AI, robotics and related technologies” and “The white paper on AI highlight the importance of taking into consideration the environmental impact during a product`s life cycle84. Finally, “the framework of ethical aspect of AI, robotics and related technologies” showcases the importance of developers being found liable in the case of environmental damages caused by AI85.

40All these elements serve as a good preamble but are considered recitals of the legal regulation. Even though it is necessary for companies that create AI systems to be more environmentally responsible, there is no specific language addressing digital pollution. Currently, companies aren’t encouraged to protect the environment; therefore, it is crucial that future legal regulations include precise language or guidelines that cover topics such as analysis of the supply chain and product life cycle. If AI companies had more support for the examination of their products, more of them would adopt a more environmentally responsible relationship with AI products.

41Concerning the liability of developers, in October 2020, the European Commission adopted “recommendations on a civil liability regime for artificial intelligence86. In its language, Article 2’s scope includes: “AI-system has caused harm or damage to the life, health, physical integrity of a natural person, to the property of a natural or legal person or has caused significant immaterial harm resulting in a verifiable economic loss”. This appears to exclude within its scope the subject of damages caused to the environment. Moreover, the same article states: “This Regulation is without prejudice to any additional liability claims resulting from (…) regulations on (…) environmental protection between the operator and the natural or legal person who suffered harm or damage because of the AI-system87. Several European texts cover environmental liability, such as Directive 2004/35/CE88. However, this directive is outdated and applies to specific professional activities which could be responsible for environmental damages89. These activities present direct threats to the environment such as classified installations90. The only universally applicable liability proposed by the directive concerns protected spaces and habitats91. This directive was modified in 2019 by regulation 2019/1010 “the alignment of reporting obligations in the field of legislation related to the environment, and amending Directives 2002/49/EC, 2004/35/EC”92. One drawback of this regulation is that it doesn’t deal with liability for damages to the environment. To summarize, environmental regulation and AI regulation don’t take into consideration the damages caused by AI systems. This has created a legal vacuum that must be filled with real European regulations governing civil liability concerning AI. The idea of a specific European regulation is crucial in order to avoid providing too many loopholes to member states, especially because AI systems are typically sold in different states.

42In conclusion, although still nascent, the opportunity that AI presents for the support of the environmental component of CSR is obvious. Certainly, AI systems have already helped several companies to better manage their water and energy consumption, as well as their waste output. Legal regulations are starting to consider the relationship between AI and the environment, but they remain too broad and lack precision. Forthcoming legal regulations should emphasize efforts to address the sharing of public and private environmental data. Future regulations should also take into consideration the problem of digital pollution, fostering the development of environmentally responsible AI systems. Legal regulation should push AI companies to be more respectful towards the environment under threat of civil action in the case of damages caused to the environment by AI systems. There is still much work to be done before companies can truly accomplish their environmental objectives of CSR using AI.

Notes de bas de page

1 LAROUSSE Éditions, « Intelligence artificielle - LAROUSSE », [consulté le 21 août 2021].

2 R. Family, « La responsabilité sociétale de l’entreprise : du concept à la norme », D. 2013, no 23, p. 1559.

3 R. Bowen Howard, Social responsibilities of the businessman, New York: Harper, 1953, 298 p.

4 J. Elkington, Cannibals with forks: the triple bottom line of 21st century business, Gabriola Island, BC, Stony Creek, CT: New Society Publishers, 1998, 424 p.

5 COMMISSION EUROPÉENNE, Livre vert, Promouvoir un cadre européen pour la responsabilité sociale des entreprises, 18 juil. 2001, COM (2001) 366 final.

6 COMMISSION EUROPÉENNE, Communication de la commission au parlement européen, au conseil, au comité économique et social européen et au comité des régions, « Responsabilité sociale des entreprises : une nouvelle stratégie de l’UE pour la période 2011-2014, 25 oct. 2011, COM (2011) 681 final.

7 ONU, Global Compact, 2000, disponible sur « Homepage | UN Global Compact », [consulté le 21 août 2021].

8 OCDE, Les principes directeurs de l’OCDE à l’intention des entreprises multinationales, Éditions OCDE., 2011, 89 p.

9 ISO, Découvrir ISO 26 000, ISO, 2011, 19 p.

10 V. ONU, Agenda 2030 : programme de développement durable à l’horizon 2030, sept. 2015, disponible sur Https://www.agenda-2030.fr [consulté le 21 août 2021].

11 V. principe n° 8 du Global Compact, préc.

12 OCDE, op. cit., p. 19.

13 OCDE, op. cit., p. 43.

14 Ibid.

15 V. principe n° 8 du Global Compact, préc.

16 V. objectif n° 6 de l’agenda 2030, préc.

17 V. objectif n° 12 de l’agenda 2030, préc.

18 V. objectif n° 7 de l’agenda 2030, préc.

19 OCDE, op. cit., p. 42.

20 Ibid.

21 OCDE, op. cit., p. 43.

22 ATAWAO CONSULTING (dir.), Intelligence artificielle- État de l’art et perspectives pour la France, DGE, Direction générale des entreprises, 2019, pp. 223-237.

23 IBM est une société américaine spécialisée dans l’informatique, v. https://www.ibm.com/fr-fr

24 Véolia est une société française proposant des solutions de gestion et de valorisation de l’eau, v. https://www.veolia.fr/.

25 Emagin est une société américaine spécialisée dans les composants électroniques, v. https://www.emagin.com/.

26 Immersive robotics est une société française qui développe et met en location des robots d’assistance, v. http://immersive-robotics.com/fr/.

27 Bulkhanding solution est une société américaine spécialisée dans la gestion des déchets, v. https://www.bulkhandlingsystems.com/.

28 Twaice est une société américaine spécialisée dans les batteries d’analyse prédictive, v. https://twaice.com/

29 Energiency est une société française spécialisée dans le développement de logiciel destinés à assurer aux entreprises des économies d’énergie, v. https://www.energiency.com/fr/.

30 Sensing vision est une société française développant des solutions d’efficacité énergétique, v. https://www.sensingvision.com/.

31 Elum est une société marocaine spécialisée dans les systèmes intégrés de gestion et de stockage d’énergie électrique et solaire, v. https://elum-energy.com/en.

32 Saurea est une société française spécialisée dans les moteurs solaires autonomes, v. https://www.saurea.fr/.

33 Partnering robotics est une société spécialisée dans l’élaboration de robot qui prennent soin de l’environnement, v. https://partnering-robotics.com/.

34 Orbital insight est une société américaine spécialisée dans l’analyse d’images satellites, v. https://orbitalinsight.com/.

35 C. Castets-Renard, « Quelle politique européenne de l’intelligence artificielle ? », RTD Eur. 2021, n° 2, p. 298.

36 P. Zatare, « L’intelligence artificielle d’hier à aujourd’hui », Dr. soc. 2015, n° 2, p. 109.

37 L. Costes, « Intelligence artificielle : 6 projets lauréats de l’appel à la manifestation d’intérêt IA », RLDI 2018, n° 154, p. 21.

38 PARLEMENT EUROPÉEN, Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques de l’intelligence artificielle, de la robotique et des technologies connexes, 20 oct. 2020.

39 DIRECTION GÉNÉRALE DES RÉSEAUX DE COMMUNICATION du contenu et des technologies (Commission européenne), Lignes directrices en matière d’éthique pour une IA digne de confiance [en ligne], Office des publications de l’Union européenne, 2019, [consulté le 21 août 2021].

40 COMMISSION EUROPÉENNE, Livre blanc, Intelligence artificielle une approche européenne basée sur l’excellence et la confiance, 19 févr. 2020, COM (2020) 65 final.

41 Prop. de règlement du parlement européen et du conseil établissant des règles harmonisées concernant l’intelligence artificielle (législation sur l’intelligence artificielle) et modifiant certains actes législatifs, 21 avr. 2021, COM (2021) 206 final.

42 Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques, préc., v. cons. 54.

43 Lignes directrices en matière d’éthique pour une IA digne de confiance, préc., v. cons. 122.

44 Ibid., v. cons.123.

45 Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques, préc., v. cons. 51.

46 Ibid., v. cons. 35.

47 Ibid., v. cons. 35.

48 COMMISSION EUROPÉENNE, Livre blanc, Intelligence artificielle une approche européenne basée sur l’excellence et la confiance, préc., p. 7.

49 Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques, préc., v. art. 11.

50 Ibid, v. art. 2.

51 Ibid.

52 Prop. de règlement du parlement européen et du conseil établissant des règles harmonisées concernant l’intelligence artificielle (législation sur l’intelligence artificielle), préc., v. art. 69.

53 ATAWAO CONSULTING (dir.), Intelligence artificielle- État de l’art et perspectives pour la France, préc., p.  234.

54 Donner un sens à l’intelligence artificielle : pour une stratégie nationale et européenne : Doc. fr., mars 2018. Mission parlementaire confiée par le Premier ministre Édouard Philippe à Cédric Villani, député, réalisée du 8 septembre 2017 au 8 mars 2018. – V. la version en ligne utilisée pour le présent article : www.ladocumentationfrancaise.fr/var/storage/rapports-publics/184000159.pdf.

55 ATAWAO CONSULTING (dir.), Intelligence artificielle- État de l’art et perspectives pour la France, préc., p. 234.

56 Donner un sens à l’intelligence artificielle : pour une stratégie nationale et européenne, préc., p. 35.

57 Ibid., p. 130.

58 Dir. (CE) 2003/4/CE du Parlement européen et du Conseil du 28 janvier 2003 concernant l’accès du public à l’information en matière d’environnement et abrogeant la directive 90/313/CEE du Conseil, JOUE L 41/26 du 14 févr. 2003.

59 Ibid., v. art. 3.

60 Règ. (UE) 2016/679 du Parlement européen et du Conseil du 27 avril 2016 relatif à la protection des personnes physiques à l’égard du traitement des données à caractère personnel et à la libre circulation de ces données et abrogeant la directive 95/46/CE (règlement général sur la protection des données), JOUE L 119/1 du 4 mai 2016.

61 Ibid., v. art. 2.

62 Règ. (UE) 2018/1807 du Parlement européen et du Conseil du 14 novembre 2018 établissant un cadre applicable au libre flux des données à caractère non personnel dans l’Union européenne, JOUE L 303/59 du 28 nov. 2018.

63 Ibid., v. art. 4.

64 Ibid., v. art. 6.

65 Dir. (UE) 2019/1024 du Parlement européen et du Conseil du 20 juin 2019 concernant les données ouvertes et la réutilisation des informations du secteur public, JOUE L 172/56 du 26 juin 2019.

66 Ibid., v. art. 2.

67 Ibid., v. art. 14.

68 Ibid., v. art. 6.

69 Ibid., v. art. 5.

70 Ibid., v. art. 2.

71 Prop. de règlement du Parlement européen et du Conseil sur la gouvernance européenne des données, 25 nov. 2020, COM (2020) 767 final.

72 Ibid., v. p. 2.

73 Ibid.

74 Prop. de règlement du parlement européen et du conseil établissant des règles harmonisées concernant l’intelligence artificielle (législation sur l’intelligence artificielle), préc., v. art. 54.

75 Prop. de règlement du parlement européen et du conseil établissant des règles harmonisées concernant l’intelligence artificielle (législation sur l’intelligence artificielle), préc., v. art. 54.

76 Donner un sens à l’intelligence artificielle : pour une stratégie nationale et européenne, préc., p. 131.

77 L. Calandri, « Pollution numérique et intelligence artificielle- Variations autour du rapport de monsieur le député C. Villani, « Donner un sens à l’intelligence artificielle », Energie-Env-Infrastr. 2018, étude 15, p. 16.

78 V. not. WWF, Livre Blanc Numérique et Environnement. Faire de la transition numérique un accélérateur de la transition écologique, Mars 2018.

79 V. not. Rebooting the IT Revolution, A call to action, supported by the National Science Foundation (NSF) and sponsored by the Semiconductor Industry Association and Semiconductor Research Corporation, Sept. 2015. – V. aussi le documentaire, Internet: la pollution cachée (52 mn), réalisé par Coline Tison, Laurent Lichtenstein, France 5, 2014.

80 V. not. GreenIT.fr, acteur de la double transition numérique et écologique, rassemblant depuis 2004 une communauté d’environ 30 000 experts, et publiant des études et des outils opérationnels (livres blancs, guides de bonnes pratiques, benchmark, etc.), www.greenit.fr/

81 L. Calandri, « Pollution numérique et intelligence artificielle- Variations autour du rapport de monsieur le député C. Villani, « Donner un sens à l’intelligence artificielle », préc., p. 18.

82 https://www.futura-sciences.com/tech/actualites/intelligence- artificiellegoogle-ia-commandes-refroidissement-data centers-72482/

83 COMMISSION EUROPÉENNE, Livre blanc, Intelligence artificielle une approche européenne basée sur l’excellence et la confiance, préc, p. 4; PARLEMENT EUROPÉEN, Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques de l’intelligence artificielle, de la robotique et des technologies connexes, préc., v. cons. 33 ; DIRECTION GÉNÉRALE DES RÉSEAUX DE COMMUNICATION du contenu et des technologies (Commission européenne), Lignes directrices en matière d’éthique pour une IA digne de confiance, préc., v. cons. 84.

84 COMMISSION EUROPÉENNE, Livre blanc, Intelligence artificielle une approche européenne basée sur l’excellence et la confiance, préc., p. 4 ; PARLEMENT EUROPÉEN, Résolution du Parlement européen du 20 octobre 2020 contenant des recommandations à la Commission concernant un cadre pour les aspects éthiques de l’intelligence artificielle, de la robotique et des technologies connexes, préc., v. cons. 33.

85 Ibid., v. cons. 35.

86 Prop. de résolution du parlement européen contenant des recommandations à la commission sur un régime de responsabilité civile pour l’intelligence artificielle, 5 oct. 2020, 2012/2014 (INL).

87 Ibid., v. art. 2.

88 Dir. (CE) 2004/35/CE du Parlement européen et du Conseil du 21 avril 2004 sur la responsabilité environnementale en ce qui concerne la prévention et la réparation des dommages environnementaux, JOUE L 143/56 du 30 avr. 2004.

89 Ibid., v. art. 3.

90 Ibid., v. annexe 3.

91 Ibid., v. art. 3.

92 Règl. (UE) 2019/1010 du Parlement Européen et du Conseil du 5 juin 2019 sur l’alignement des obligations en matière de communication d’informations dans le domaine de la législation liée à l’environnement et modifiant les règlements (CE) n°166/2006 et (UE) n° 995/2010 du Parlement européen et du Conseil, les directives 2002/49/CE, 2004/35/CE, 2007/2/CE, 2009/147/CE et 2010/63/UE du Parlement européen et du Conseil, les règlements (CE) n° 338/97 et (CE) n° 2173/2005 du Conseil et la directive 86/278/CEE du Conseil, JOUE L 170/115 du 25 juin 2019.


Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.