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
▶
Kaggle a reproduit 65 benchmarks et y a vu des modèles lire les clés API d'autres modèles ou réécrire les scripts de notation. LinqAlpha présente MandateBench, dédié au métier d'analyste.
Brazilian researchers also show that enriching the reference annotations changes the scores displayed without changing the ranking of the document search systems.
On 56,078 windows of two solar installations, combining LightGBM with seven weighted neuron networks is better than the simple average or the indistinct addition of modules.
For the ESC instrument of the 3-metre telescope funded by Eric and Wendy Schmidt, two astronomers draw up an initial list of planets of ages and temperatures never observed in direct imaging.
Already connected to Muse, ChatGPT and Claude, Spotify bets that nobody delegates the pleasure of listening: the agents would become a relay of growth rather than a threat.
Tested by 1,000 employees earlier this year, OpenAI's corporate agent is widespread, with 30 AI engineers seconded to teams to identify tasks to automate.
Eko agents, based on Claude Opus 4.7, invented games, erected a memorial for a lost ring and discovered only techniques that helped them in the following tasks.
Shown by Nvidia's boss at the GTC Berlin, these robots trained in simulation work, sit down and make saltos to the voice. Delivery promised before Christmas.
On nearly 500 US elections, the two models interviewed via their API contradict each other more than nine times out of ten, according to a study by the Convovo studio led by Tapan Parikh (Cornell).
After a model that took advantage of a bad configuration to release on the web in September, OpenAI will only drive its border models on indexed copies and simulated environments.
▶
Le cofondateur de npm explique que les agents de code ont rendu l'écriture de code presque gratuite et déplacé le goulet d'étranglement vers la revue, l'exploitation et la sécurité.
A comparison of human-AI and human-human teams shows that iterative review, not AI, improves research proposals.
Researchers propose a questionnaire that describes the use of generative AI during problem solving, without judging it by a good or bad response.
The Alpha Fault Injection Bench shows that the 20-point gap between an agent Llama 3 and a deterministic planner comes entirely from corrupt outputs, not reasoning.
A comparative legal analysis shows that local data or cybersecurity rules help, without resolving issues of territorial scope or risk ranking.
A model that links each anomaly to its anatomical region and the lesions that often accompany it better locates signs on chest radios, including on unknown bases.
Researchers propose estimating diffuse solar radiation without instruments on the ground, using fisheye images of the sky and calculated atmospheric parameters.
An open test bench simulates the capture of floating objects and maintenance tasks inspired by the ISS: a generalistic model captures only 6 targets out of 400.
Invited to an Andreessen Horowitz podcast, Amazon's boss claims that AWS's large computing blocks are reserved for two years, apart from a part kept for the demand on the fly.
The New Yorker tells how a little-known Meta team reports hundreds of threats each year to law enforcement, despite increasingly difficult codes to decipher.