Skip to main content

AI for Humanity: EDF, Thales and Total open the first joint industrial laboratory for Artificial Intelligence

  • In line with the manifesto on Artificial Intelligence for Industry signed by 14 leading global industry players, EDF, Thales and Total today launched their first joint research laboratory for Artificial Intelligence (AI).
  • As the first step in the companies' AI action plan, this laboratory located at the EDF Lab Paris-Saclay (Palaiseau) will develop trusted AI for critical systems and support closer cooperation between the three groups at Saclay.
  • The laboratory will pool the research efforts of EDF, Thales and Total in the priority areas of AI explainability, reinforcement learning and the combination of AI and simulation for industrial problem-solving.

The manifesto on Artificial Intelligence for Industry, signed on 3 July by Air Liquide, Dassault Aviation, EDF, Renault, Safran, Thales, Total and Valeo, and later by Schlumberger, ST, Orange, Naval Group, Michelin and Saint Gobain, reflects the shared objective of leveraging AI to drive growth and create jobs in industry in line with the values and vision of France's national #AIforhumanity strategy.

This new research laboratory for Artificial Intelligence, co-directed by EDF, Thales and Total, marks a first step in making the industry manifesto a reality.

Pooling research efforts to develop Artificial Intelligence for critical systems

In creating this first joint laboratory, EDF, Thales and Total aim to develop AI technologies to meet the needs of critical industrial systems.

Critical systems on board aircraft or at energy production facilities require the highest level of reliability and call for the development of trusted, explainable and certifiable AI, which is the core objective of the new research laboratory.

The laboratory will also share resources, data and expertise to develop advanced AI technologies that will benefit all three partners' activities by optimising tasks, offering faster access to relevant information, and improving industrial performance, energy efficiency and industrial facility availability.

Explainability, learning and simulation: three keys to trusted AI

By working on explainability, the partners will seek to ensure that an automated system is able to explain the reasoning that led to its proposed response. This is key to the trust that users place in AI, particularly in critical systems.

Reinforcement learning involves teaching a system a behaviour by positively or negatively rewarding the AI as it learns.

Finally, by using AI combined with simulation, experts can generate interactions between AI and a physical model in order to optimise a system's behaviour within a set of parameters. In this way, AI can significantly improve the configuration settings of complex physical systems, which would be much more time-consuming, and in some cases impossible, with traditional optimisation methods.

Each of these three areas of research is expected to provide valuable insights to engineers and strengthen their decision-making ability, helping to ensure that AI serves humanity by improving the quality of human decisions.

Open innovation to tackle key AI challenges

Located within the EDF Lab in the Saclay research and technology cluster south of Paris, the laboratory will benefit from its close geographic proximity to the partners' existing research teams; and as part of one of the world's largest academic and innovation ecosystems, it is expected to attract doctoral candidates from the region's top academic research institutes.

Governed by a research agreement signed by EDF R&D Director Bernard Salha, Total R&D Director Marie-Noëlle Semeria and Thales Scientific Director Marko Erman, the laboratory aims to step up to the key Artificial Intelligence challenges launched by France's National Innovation Council.

For more information about AI for Humanity: https://www.aiforhumanity.fr/

 

Contact
Alice Pruvot, Head of Media Relations, Aeronautics & Defense
+33 7 70 27 11 37 alice.pruvot@thalesgroup.com