Logo
Thématique
Economie, emploi et travail
Politiques, institutions et démocratie
Libertés publiques et éthique
Inclusion numérique et solidarité

Fichier 1

Événement

Can AI get close to "human intelligence"?

  • Libertés publiques et éthique
  • Aux Sources du Numérique
23 février 2021

Can AI get close to "human intelligence"?

  • Libertés publiques et éthique
23 février 2021
17h30 à 19h
The debate was streamed live / https://www.youtube.com/watch?v=70gjYwW10dU

Can AI get close to "human intelligence"?

Over the past two decades, artificial intelligence (AI) has been improved by many technological advances, mainly based on the collection and use of big data, and algorithms relying on a method called "deep learning". As a result, AI can now be applied to many common processes and tasks, such as speech recognition, image labelling, playing board games online or text translation. However, these technologies cannot yet function as autonomous AI solutions, and are thus unable to perform many complex operations like human beings. 

In their book, Rebooting AI: building Artificial Intelligence we can trust (Pantheon Books, 2019), Professors Gary Marcus and Ernest Davis review AI's current achievements. They also discuss the potential for a new generation of AI, one that could support every aspect of our lives - from making medical decisions to driverless cars. 

The authors argue that the central problem with current AI is that it is too "narrow", with solutions programmed for particular tasks based on fixed rules. As a result, those technologies are not able to cope with new experiences they haven’t been programmed to recognise, and cannot respond to real-world situations that feature any degree of unpredictability or creativity. 

Taking AI to the next level will thus require developing machines that can adapt more easily and quickly to different situations. As argued by Marcus and Davis, this future AI could be more reliable and more trustworthy, as it would have a deeper understanding of the world and of human values.

But how can we possibly create machines with such broad intelligence that they can deal with problems they have never been confronted with before? And more importantly: what would be the consequences for “life” as we know it?

This debate in presence of Professor Ernest Davis was held and broadcasted live on YouTube on Tuesday, February 23 at 5:30 pm (Paris local time), and was the 40th edition of our meeting series "Aux Sources du Numérique". 

 

Informations pratiques

23 février 2021
17h30 à 19h
The debate was streamed live / https://www.youtube.com/watch?v=70gjYwW10dU

Les intervenants

Ernest Davis
Professor of Computer Science at New York University
Henri Isaac
Président de Renaissance Numérique, Maître de conférences, PSL Université Paris-Dauphine
Nicolas Vanbremeersch
Directeur et fondateur de Spintank

Nos soutiens

Logo spintank