Passer au playerPasser au contenu principal
  • il y a 1 jour
Chacun son domaine. Le pape Léon XIV s’est entouré d’éminents théologiens catholiques et de l’un des cofondateurs de la start-up américaine Anthropic pour présenter, ce lundi 25 mai, son premier texte majeur consacré à « désarmer » l’intelligence artificielle pour « l’empêcher de dominer l’humain ». Dans Magnifica Humanitas (Humanité magnifique), publié par le Vatican, il dénonce aussi les « nouvelles formes d’esclavage » derrière l’essor fulgurant de l’IA.

Christopher Olah, le cofondateur d’Anthropic, a déclaré que sa présence à la cérémonie s’expliquait par l’ampleur des enjeux liés à l’IA, dans un message partagé lundi sur X : « Les questions soulevées par l’IA dépassent le cadre de la communauté de l’IA. Nous avons un besoin urgent que le monde entier — les religions, la société civile, les universitaires, les gouvernements — participe à la mise en place d’une issue positive. »

Catégorie

🗞
News
Transcription
00:00Artificial intelligence needs to be disarmed.
00:15Artificial intelligence already touches many areas of our lives
00:20and affects decisions that shape human coexistence.
00:23It is also dramatically changing how war is waged.
00:30Other very troubling voices have also reached me
00:33about increasingly autonomous weapons systems,
00:37practically beyond any human reach to govern them effectively.
00:41I hear very troubling accounts of algorithms
00:44that can block access to health care, employment, and security
00:49on the basis of data tainted by prejudice and injustice.
00:54Artificial intelligence needs to be disarmed.
01:00The word is strong, I know,
01:02but deliberately chosen because this moment needs words
01:06capable of attracting attention, awakening consciences,
01:10and indicating paths forward for humanity.
01:14Let's not fear artificial intelligence,
01:17but constantly keep the question of the human in play.
01:22We cannot be careless with our most powerful technical instruments.
01:32In a special way, I'd like to thank Mr. Ola for accepting our invitation.
01:39In turn, in the name of the Church,
01:42I accept your invitation to walk together,
01:45to listen and to speak,
01:47and together to find the way for humanity
01:50in this time of artificial intelligence.
02:00Every Frontier AI lab, including Anthropik,
02:03operates inside a set of incentives.
02:07and constraints that can sometimes conflict
02:10with doing the right thing.
02:13The pressure to stay commercially viable
02:16and to stay at the frontier of research.
02:20Geopolitical pressure
02:22and the older, plainer pressures
02:24of pride and ambition.
02:27AI development is concentrated
02:29in a handful of wealthy nations.
02:32How will we ensure that the gains of AI
02:34are shared globally?
02:34It is an unsolved problem.
02:36And it is the kind of problem
02:38the Church has historically refused
02:40to let the world ignore.
02:41And I will be honest,
02:42we keep finding things that are mysterious,
02:45even unsettling.
02:46We find structures that mirror results
02:48from human neuroscience.
02:49We find evidence of introspection.
02:51We find internal states that,
02:53functionally, mirror joy, satisfaction,
02:55fear, grief, and unease.
02:58I don't know what that means,
02:59but I think it warrants ongoing discernment.
03:02really cool says it.
03:03Thank you very much.
03:04Thank you.
03:04Thank you everyone for that.
03:08Thank you.
03:08Thank you.
03:09Sous-titrage Société Radio-Canada
Commentaires

Recommandations