Google Brings Gemini Voice AI to Apple CarPlay
Google is rolling out its Gemini Live AI assistant to Apple CarPlay, enabling hands-free conversations for iOS users during commutes, building on its existing Android Auto integration....
Read article →
Science & Applications Correspondent
Owen Mercer follows the practical uses of machine learning in science and everyday systems. His coverage centers on usefulness: who benefits, what has been demonstrated and what remains a prototype. He writes patient explanations of technically demanding subjects.
1,228 published articles · Showing 1–20
Google is rolling out its Gemini Live AI assistant to Apple CarPlay, enabling hands-free conversations for iOS users during commutes, building on its existing Android Auto integration....
Read article →An X user's unverified claim that Microsoft leaked classified CoPilot information to China has not been corroborated by any independent source or official statement....
Read article →Palantir's hiring of former Sen. Jon Tester as an adviser comes as the company prepares for increased oversight from a Democratic-controlled Congress, particularly over its work with the Department of Homeland Security....
Read article →CMU's ServerBench evaluates coding agents on designing LLM serving systems under real workloads, with Claude Sonnet 5.5 achieving 4.87x the vLLM baseline on Gemma E4B but only 22 of 33 agent-built systems passing validation....
Read article →An X user's claim that Anthropic's Claude Code leaked an internal multiplayer UI, the day after an engineer said it wasn't available, has not been independently verified and Anthropic has not commented....
Read article →Cities generate large amounts of data from cameras, sensors and operational systems, but turning it into useful action remains a challenge, NVIDIA said in a newsroom post. The company's agentic AI and physical AI technologies aim to close that gap....
Read article →OpenAI’s account describes Ultrafast as an internal incident-response tool that searches across several possible causes and data sources. The company also cited employee use of Codex for planning, browser tasks and refactoring....
Read article →The company has named a proposed ticker but has not set the offering size or price range. The financial figures circulating on social media could not be verified against the filing....
Read article →The tools target developers building models for robots and autonomous vehicles, where collecting examples of rare or hazardous events can require costly physical tests....
Read article →Stack Overflow reports a significant increase in user acquisition and agent creation after integrating its agentic product into ChatGPT. The company also credited OpenAI's Codex for reducing connector development time from weeks to days....
Read article →The FBS Debate protocol improves scalable oversight by ensuring that AI honesty is a dominant strategy. It provides worst-case correctness guarantees for problems with stable decompositions....
Read article →Google DeepMind's Antigravity platform integrates AI agents and a new protein language model to identify potential antibiotics. One identified candidate for respiratory infections is currently undergoing laboratory validation....
Read article →Apertus 1.5 introduces multimodal capabilities and a 262k token context window to the 8B and 70B model families. The release includes open weights and data, with initial deployments in Proton Lumo and Infomaniak Euria....
Read article →The model allows life sciences teams to reason across biological data and coordinate workflows across scientific tools, databases, and literature. These capability descriptions are provided by OpenAI and have not been independently tested....
Read article →LangChain published guidance on using its LangSmith platform for agent evaluation, describing workflows that combine offline datasets and online production monitoring....
Read article →The proposed BSIPipe method uses bidirectional interleaved pipelines to improve LLM training throughput. Experiments on GPT and LLaMA3 show speedups of up to 1.13x over existing baselines....
Read article →NVIDIA has introduced the cuPotential library and expanded its CUDA-X portfolio to help semiconductor manufacturers use AI agents for atomic simulation and fab optimization....
Read article →NVIDIA and Cisco are developing the USFP optical module to increase bandwidth density in AI data centers while reducing power consumption. The new standard targets 400Gb/s-per-lane signals to support larger AI clusters without adding network layers....
Read article →Microsoft plans to transition its AI inference workloads to in-house Maia chips by 2027 to lower costs and decrease dependency on Nvidia and AMD GPUs....
Read article →A new preprint describes UniConFlow, a flow-matching framework for robot motion planning that handles multiple constraints. It reports outperforming baselines across three experiments, but the findings are unreviewed claims without independent validation....
Read article →