Nvidia launches free tool to pool idle computers for AI
PAIR links compatible home PCs and uses them for local AI inference when they are idle.
Models, agents, applications: artificial intelligence put to work on real problems.
Here, artificial intelligence is judged on the evidence: what it does, for whom, at what cost. The section follows models and their measurable progress, agents doing real work, and above all deployments — a hospital screening more patients, a power grid better balanced, a public service answering faster. AI-as-spectacle, productless funding rounds and prophecy get no space.
We favour documented use cases with the numbers attached: how many examinations, what error rate, what savings. Non-Western players — Chinese, Indian, Emirati labs — are covered on the same footing as American ones, because much of the real deployment happens there. Every article cites its sources and separates the prototype from the service in production.
PAIR links compatible home PCs and uses them for local AI inference when they are idle.
NYU Tandon and FDNY trained an AI model on nearly 1,000 ambulance responses in 2023.
Starting with simple trials helped the networks learn a complex go/no-go timing task more accurately.
The “Custom AI Behaviors” beta builds complex if-then rules from natural language.
The launch connects smart hardware, industry applications and developers through WorkBuddy’s agent capabilities.
Built for retrieval-augmented systems, TOHA needs no separately trained detector.
CW-Net turns opaque driving-model reasoning into concepts such as “close to cyclist.”
The records came from smart meters, building management systems and industrial grids.
Researchers propose a framework for AI agents that use internal states to interpret surroundings and choose actions.
Through the MCP server, AI clients can now edit project data within their existing permissions.
Alexa for Shopping checks sender details, content, timing and metadata against Amazon’s message history.
EPFL's GOLLuM pairs a language model with a Gaussian process that estimates when its suggestions may be wrong.
The BPI-AI2N uses Renesas’ RZ/V2N and reaches up to 15 TOPS on sparse computations.
Thomas Jefferson’s framework searched records from breast and lung cancer patients for treatment-related heart damage.
The generally available model improves from 24.7% to 52.6% on Terminal-Bench-Science.
Mifeng, JD.com and U.S.-based Figure are expanding embodied-intelligence data collection into communities, homes and real-world workplaces.
Detection results will be sent directly to customers’ security teams, eliminating the need for manual review by Anthropic.
Fable 5.1 is available today through cloud platforms and Anthropic’s API.