Can we foster an ethical adoption of AI?

Can we foster an ethical adoption of AI? This was the overarching question of the conference organised by the Ville d’Issy-les-Moulineaux and the GlobalAI Association where I spoke last Friday in a panel with Eric Legale, Emmanuel R. Goffi, Delphine Dogot, Michel-Marie MAUDET, and Bertrand CASSAR. The short answer? Yes, but only if we build AI applications that can change reality in the way we want, and not the other way around. If we build “democratic” AI.

As for the long answer…in his keynote speech, Prof. Goffi, rightly pointed out that ethics is not just about preventing bad things from happening, about imposing laws. Legislation is crucial and requires painstaking cooperation among governments and companies, as highlighted out by Prof. Dogot. But beyond preventing the bad things, we need to think about what good AI means.

To me, good AI is “democratic”, not in a political but in a literal sense, AI as “power of the people”. That is, building AI models from the ground-up that reflect the needs and aspirations of the workers or citizens that will use them, as opposed to the views of a few AI wizards, as opposed to an “Autocratic” AI, if you will.

This is where Michel-Marie, Bertrand and I came in. Michel spoke about the open source model Lucie, Bertrand about the AI strategy at La Poste. I discussed a few examples of private language models Altruistic has built, for Ferrari in Italy and the Government of Canada.

Thank you to the GlobalAI Association, Daniela Leveratto, Matteo Satta and Marco Bena for the invitation. And to Emmanuel Dupuy for the effective moderation.