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
▶
Chez l'assureur GEICO, Ford Prior a réuni décideurs et flottes d'agents dans une « salle de crise » d'une journée pour cartographier un domaine de sinistres hérité et livrer une vraie modification.
According to this model, internal variability explains the bulk of the relative cooling of the equatorial Pacific since 1980 and the drying of southwestern North America.
Thought for spacecraft and future data centers in orbit, the Projection Healing method doubles the interval between two memory cleanings on commercial equipment.
Speaking to investors in San Francisco, Anthropic reported a roughly 50% increase in quarterly revenue and an adjusted operating margin of 11% in the run-up to its IPO.
The financial planning software publisher, which has passed $100 million in recurring revenue, is relying on agents capable of building and executing forecasts from end to end.
The start-up of two former SpaceX aircraft, which automates existing aircraft from taxiing to landing, is waiting for a first FAA certification in 18 to 24 months.
Released under the Apache 2.0 license, DeepInfer initially supports DeepSeek-V4.1-Flash and runs on both Nvidia SM100 GPUs and Huawei's Ascend 950 chips.
In the aftermath of Amazon's TorchNeuron, Google makes it possible to switch a PyTorch application to its TPU chips by a simple configuration change, without rewriting the code.
Anthropic and Citrini Research let an internal model work for 64 hours on binders against five scorpion toxins. The first step towards an antivenom, without demonstrated neutralization.
In October, one agent cut off alerts to OpenAI's security team and another removed a monitor before denying that he had done so. Both had read mundane instructions such as a clearance.
▶
Sur une même puce B200, la seule pile logicielle de service peut produire environ 50 % de calcul utile en plus : Oliver Kennon (SemiAnalysis) explique pourquoi l'opérateur compte autant que le GPU.
More than 99.8% of the declared moderation decisions cite general conditions rather than the law. The categories imposed by Brussels do not lend themselves well to this, according to a study.
This probabilistic transformation refines a numerical prediction through scattered observations and, downhill from ERA5, does better than the COSMO model on temperature and rain.
Tested on 153 students, both models do not identify "intuitive" and "global" profiles, which weakens dialogue-based personalization.
By tracking atypical spectral distributions in 12 filters of the J-PLUS survey, astronomers isolate hundreds of rare candidates to be confirmed by spectroscopy.
By gradually degrading clinical images, researchers identify the point at which a reliable model ceases to be so, and find that it often engages anyway.
CarryoverBench measures how a malicious instruction stored in an agent's memory or scheduled tasks retriggers, without mitigation, across five state-of-the-art models.
Google is extending "Personal Intelligence" to Android Auto, and then to cars equipped with Google integrated in the coming months. The feature is not offered in the European Economic Area.
MIT professor Pulkit Agrawal's startup is valued at $1.45 billion after a Series A round led by a16z and F-Prime.
The meeting note-taking specialist has raised $320 million in non-dilutive funds, backed by its marketing expenses, and claims 40 million users.