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
▶
Deux spécialistes de New York Life ont analysé 1 368 incidents d'IA publics pour en tirer une méthode d'évaluation : les contrôles de déploiement défaillants sont les plus liés aux dommages.
This system continuously updates a robot policy based on the interventions of an operator and his own tests, without criticism learned or reward model.
By combining three types of on-board sensors on five geostationary satellites, this deep learning system aims to provide reliable rain data in time for flood warning.
Astronomers have applied AdaCLIP, a "zero-shot" fault detection model designed outside astronomy, to the CERS and NGDEEP surveys to identify unusual morphologies.
Out of three cases of Korean clinical examination and 7,480 double-blind evaluations, three out of four models remain slightly below the human expert; Llama 3.1 70B picks up on all criteria.
A study of 23 Indian students and young graduates describes a large gap: AI-oriented companies, but exams and campus hiring tests that exclude it.
When a code agent compacts its memory, pi, opencode, Cline, Help and OpenHands erased that said: a forged approval survives in 87 to 100% of cases, compared to 0% at Codex or Gemini CLI.
Specializing in the deployment of custom AI systems for the world's largest groups, AuxoAI has completed a series A led by NTT, also an Anthropic investor.
Huawei provides Ascend chips and software stack, DeepSeek workloads and optimization: According to SemiAnalysis, neither of them would do it alone, but the Ascend ecosystem does not benefit.
In Foreign Affairs, historian Niall Ferguson, former CEO of Google Eric Schmidt and researcher Lisa Klaassen compare China-American rivalry on AI with empire history.
▶
Sarah Chieng (Cerebras) et Cova Lee (OpenAI) détaillent la chaîne d'agents qu'ils utilisent pour produire leurs contenus, de l'idée à la publication, en laissant à l'IA les tâches fastidieuses.
Out of 8,524 Brazilian patients, images of a Nikon retinograph were twice as often inoperable as those of a Canon. A study shows that this gap has an uneven impact on 12 AI diagnostic tasks.
A single control policy, trained in several stages and limited to embedded sensors, allows precise point-to-point jumps in simulation, and a first vertical jump was made on a real robot.
A study compares Gaussian processes, recurring networks and a model with normalizing waves, FloTS, to anticipate the image quality of the sky, useful for telescopes as well as optical links in free space.
By surveying the internal states of GraphCast on 205 cyclones, a team shows that each physical attribute is distributed over several characteristics, according to an organization that repeats itself from layer to layer.
During GM Forward, the manufacturer introduced a voice assistant capable of driving the vehicle's functions, promised the organization of the daily, and announced the extension of his "eyes-off" driving.
In presenting the third quarter results, Mark Zuckerberg claimed more than 10 million users for his agent Muse, more than a quarter of whom made purchases there, and promises to monetize faster than usual.
After AWS in the spring, it's Google Cloud's turn to host GPT models. OpenAI is now present at the top three cloud providers, less than six months after Microsoft's exclusivity ended.
While Amazon blocks the purchasing agents, Walmart takes them to the red carpet: an MCP server, agent profiles with validation or spending limits, and their own conditions of use.
A honey pot from Amazon Web Services was targeted as early as October 17, while PHP's critical flaws were still under embargo. They are now being exploited on a large scale.