MIT researchers develop an efficient way to train more reliable AI agents
The technique could make AI systems better at complex tasks that involve variability.
The technique could make AI systems better at complex tasks that involve variability.
Physicists surprised to discover electrons in pentalayer graphene can exhibit fractional charge. New study suggests how this could work.
Researchers show that even the best-performing large language models don’t form a true model of the world and its rules, and can thus fail unexpectedly on similar tasks.
National Science Foundation grant expected to help New England researchers translate discoveries to commercial technology.
By emulating a magnetic field on a superconducting quantum computer, researchers can probe complex properties of materials.
System observed 8,000 light-years away may be the first direct evidence of “gentle” black hole formation.
Labs that can’t afford expensive super-resolution microscopes could use a new expansion technique to image nanoscale structures inside cells.
Models show that an unexpected reduction in human-driven emissions led to a 10 percent decline in atmospheric mercury concentrations.
Because it doesn’t need expensive energy storage for times without sunshine, the technology could provide communities with drinking water at low costs.
MIT’s innovation and entrepreneurship system helps launch water, food, and ag startups with social and economic benefits.
“Co-LLM” algorithm helps a general-purpose AI model collaborate with an expert large language model by combining the best parts of both answers, leading to more factual responses.
Physicists capture images of ultracold atoms flowing freely, without friction, in an exotic “edge state.”
Membranes based on natural silk and cellulose can remove many contaminants, including “forever chemicals” and heavy metals.
Building on a landmark algorithm, researchers propose a way to make a smaller and more noise-tolerant quantum factoring circuit for cryptography.
These zinc-air batteries, smaller than a grain of sand, could help miniscule robots sense and respond to their environment.