Until recently, Massachusetts shellfisheries relied on a low-tech system of neighbors reporting rain gauge measurements to close shellfishing areas. However, this system is inefficient and outdated. MIT researchers have developed Seashell, a web-based platform that aims to streamline natural resource management and support productivity of shellfisheries, safe human consumption, and vitality of coastal economies. The platform is scalable and can be used by state agencies and individual permit holders to make decisions in real time, ensuring reliable local food supply.
As AI accelerators evolve, understanding their capabilities is crucial for maintaining technological advantage. The Lincoln Laboratory Supercomputing Center's Lincoln AI Computing Survey (LAICS) has been tracking the development of AI accelerators since 2018, summarizing current commercial accelerators and comparing their peak performance and power. The survey aims to identify the best accelerators for specific needs.
A new workshop report from Google explores open research directions for building AI agents that users can trust. It addresses foundational privacy and security challenges faced by autonomous agents, including access to personal data and behavioral flexibility. The report aims to coordinate defenses at system, model, user, and ecosystem levels.
MIT's Transit Lab has been selected as one of 15 projects to receive $2.1 million in funding from Google.org's Impact Challenge: AI for Government Innovation. The project aims to develop a Public Transit Intelligence Hub (PTIQ) that will unify real-time monitoring, operations control, and passenger communication systems into a single AI-orchestrated platform for public transportation agencies. The goal is to improve the decision-making capabilities of transit control center staff and provide riders with more accurate information.
Instead of treating image generation as a rigid sequence of isolated steps, Diffusion Controller reframes the entire denoising process as a smooth, continuous control problem.
MIT researchers argue against concerns that algorithmic monoculture could result in systematic exclusion and other negative consequences. They systematically evaluated objections and concluded these arguments either fail or are not decisive against all forms of monoculture.
NIST convened a two-day NCST Advisory Committee meeting on Sept. 23-24, providing updates on investigations into the 2021 partial collapse of Champlain Towers South and Hurricane Maria's impact on Puerto Rico, as well as NIST's Disaster and Failure Studies Program and implementation of recommendations from a 2011 tornado investigation. The committee highlighted the importance of collaborations and extensive data collection in investigations.
A new tool developed by scientists at MIT's McGovern Institute for Brain Research uses a custom-built list of words and phrases linked to 49 suicide risk factors to estimate an individual's risk. The tool was tested in text conversations with crisis counselors and is already helping clarify which factors matter most in times of crisis. With more validation, it could be used in clinical settings and crisis-support situations.
Researchers from MIT's Senseable City Lab used machine learning to analyze traffic cameras in New York City, identifying vehicle types and estimating emissions. Visual AI can address urban planning questions, but concerns about privacy and fairness are growing. While useful, researchers caution against over-reliance on visual data.
MIT researchers have developed tiny batteries that can safely break down in the GI tract, making ingestible devices safer and more environmentally friendly. The batteries, powered by magnesium and molybdenum trioxide, can power RFID tags and stimulate ghrelin production in capsules. The technology could be applied to a range of ingestible devices for sensing and therapeutic applications.
MilleMiglia is an open-source tool that provides realistic benchmarks for optimizing complex middle-mile logistics networks. This bridges the gap between academic theory and industrial logistics, enabling researchers to improve global supply chains. Despite its importance, middle-mile logistics has received less attention in operational research due to a lack of public, high-quality data. It has applications in various supply chain scenarios, including e-commerce, retail, and time-sensitive movements.
Google Research has developed a new tool that allows educators to create custom, interactive, and guided educational simulations tailored to their objectives and curriculum. The system leverages generative user interfaces (GenUI) optimized for learning and is currently available in English for STEM subjects including physics, chemistry, biology, and math. The tool is being released to a trusted tester pool and is part of a pilot program to improve learning interactives for public use.
NRKZ Signal · Signal illustration