AI News

MIT scientists have boosted AI language models’ problem-solving skills using test-time training, setting a new record on a tough AI benchmark.

Connecty AI secures $1.8 M to revolutionize enterprise intelligence by providing data professionals with advanced AI-driven agents.

Google.org awards Karya a $1 million grant to enhance digital skills, supporting education and training initiatives across India.

Bengaluru’s AutoVRse launches VRseBuilder, an AI-powered platform that cuts VR training development time by up to 93%, enabling enterprise-wide scaling.

Anthropic is set to launch Claude 3.5 Opus soon, aiming its most powerful AI to compete with the upcoming Gemini 2.0 and other top models.

Anthropic has postponed the launch of its next AI model, Opus 3.5, without announcing a new release date. This delay highlights broader challenges in advancing large language models across the industry.

Alibaba Cloud unveils its new Qwen2.5 series, featuring the Qwen2.5-Coder-32B-Instruct model that closely aligns with human preferences.

OpenAI’s Sutskever foresees a new AI "age of discovery" as scaling large language models slows, prompting top firms to emphasize test-time compute instead.

Google DeepMind has open-sourced AlphaFold 3, empowering scientists worldwide to enhance protein modeling and accelerate drug discovery by accurately predicting molecular interactions.

Google DeepMind and Isomorphic Labs have unveiled AlphaFold 3, their newest AI-powered protein prediction tool, now available for use in scientific research.

Mozilla CEO Laura Chambers tempers AI browser excitement, predicting that meaningful AI-driven browser changes won’t emerge for another two to three years.

The US Department of Commerce has directed Taiwanese chipmaker TSMC to stop exporting advanced artificial intelligence chips to China.