AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
▶
Le RAG permet à un agent de lire, la mémoire lui permet d'écrire. Wojciech Pluta (Oracle) explique comment trier ce qui mérite d'être retenu, le confronter à l'existant, et escalader au besoin vers un humain.
Up to 24 hours of maturity, FitzRoy equals or exceeds the state of art without the blurred rain fields typical of AI, and reports record cumulations of the Texas and Missouri floods.
960 conversations of 20 simulating people in crisis show that recognizing distress and speaking with empathy is not enough to respond safely.
With identical content, the location of an injection changes everything: AI agents implicitly give more confidence to certain areas of their environment.
Out of 200 static mesh problems, thirteen models detect almost all errors, but their complacency with the student varies from 1.6% to 50%.
Gym-Anything for Robotics entrusts the creation of simulators to code agents, guided by US GDP. Average peak model score: 0.06.
Rather than looking for the rarest object, researchers ask a vision-language model what is scientifically interesting. Cost: about $5 per 10,000 images.
The new ElevenLabs sound effects model better follows instructions at several events and produces a cleaner sound, on ElevenCreative, the mobile app and the API.
Parental control, ban on catbots to minors, government tests: voters in two key states for the Senate want to frame AI, even if it hinders competitiveness.
Thrive Capital's holding company offers $10 per share for the specialist in remote surveillance, valued $15.3 billion including debt.
▶
Le cofondateur de Vals AI explique pourquoi un score unique de benchmark masque une frontière en dents de scie, et comment un modèle moins cher sur le papier peut coûter plus en consommant davantage de tokens.
The analysis of 6,150 predictive markets created by users on their own future shows persistent collective optimism, which only the individual concerned ends up correcting.
In a virtual reality simulation, 16 participants manipulated an arm mounted on rover with joystick or look. The second mode is much faster and less tiring, but less precise for removal.
Following the decline in inverter production by inverter, a multimodal model anticipates cloud passage and advances 14 reference methods, including overcast or changing skies.
Models guess "who speaks" from the content and structure of the text. Injection training only protects against half the problem; Internal probes detect the attack.
Rather than judging AI on its most publicized breakthroughs, researchers tested it on a random sample of recent conjectures. Trivial counter-examples and unoriginal evidence are alongside real novelties.
Interpreting a genetic variant requires crossing recommendations, bases and calculations. An agent who assembles executable skills significantly improves the accuracy and speed of 45 professionals.
The New York manufacturer of paillass robots, present with 11,000 machines in more than 80 countries, loops an E series led by Avego to accelerate "self-sustaining science".
Led by the German band Schwarz, Cohere's E series would be the biggest release of a Canadian startup. The federal government is involved and Cohere is committed to investing $1.7 billion in computing in Canada.
On 28 October, Kathy Hochul signed a law prohibiting chatbots from making themselves known as humans or encouraging self-harm among those under the age of 18. It shall enter into force on 1 January 2027.