Google Research
·
2 months ago
● 2 sources
Google Research released open-source tools and datasets across genomics, neuroscience, climate modeling, and healthcare to enable scientific collaboration globally. The company's suite includes DeepVariant for DNA analysis, H01 brain tissue reconstruction dataset accessed over 200,000 times, NeuralGCM for weather forecasting reaching 2 billion people, and MedGemma medical models downloaded 4.8 million times. These resources have been deployed in real-world applications such as delivering monsoon forecasts via SMS to 38 million Indian farmers and enabling genetic diagnosis in under 8 hours.
The Register
·
2 months ago
● 2 sources
Forrester predicts that by 2030, the proliferation of AI agent systems across enterprises will create uncontrolled sprawl, forcing CIOs to shift from builders to governors of AI-powered operations. The analyst warns that without proper governance structures, fragmented adoption and misaligned systems will cause systematic failures at scale by the end of the decade. CIOs will need to adopt new roles focusing on enterprise decision-making architecture, bounded autonomy patterns, and risk management to maintain control over autonomous AI agents.
Allen Institute (AI2)
·
2 months ago
Allen Institute for AI interim CEO Peter Clark discusses the organization's focus on long-term AI research and development in the open, emphasizing the risk that rapid progress may overshadow foundational work. The institute is developing open models like Olmo and Molmo, operating systems such as AutoDiscovery for cancer treatment analysis, and building tools like Asta for scientific discovery, with NSF and NVIDIA funding supporting the next generation of transparent model development. Going forward, Ai2 plans to prioritize advancing AI system reliability and transparency, creating tools for scientific research and discovery, exploring embodied AI applications, and addressing environmental and global systems challenges through open platforms.
