Study: Platforms that rank the latest LLMs can be unreliable
Removing just a tiny fraction of the crowdsourced data that informs online ranking platforms can significantly change the results.
New tissue models could help researchers develop drugs for liver disease
Two models more accurately replicate the physiology of the liver, offering a new way to test treatments for fat buildup.
How generative AI can help scientists synthesize complex materials
MIT researchers’ DiffSyn model offers recipes for synthesizing new materials, enabling faster experimentation and a shorter journey from hypothesis to use.
A portable ultrasound sensor may enable earlier detection of breast cancer
The new system could be used at home or in doctors’ offices to scan people who are at high risk for breast cancer.
A new way to “paint with light” to create radiant, color-changing items
“MorphoChrome,” developed at MIT, pairs software with a handheld device to make everyday objects iridescent.
Efficient cooling method could enable chip-based trapped-ion quantum computers
New technique could improve the scalability of trapped-ion quantum computers, an essential step toward making them practically useful.
A protein found in the GI tract can neutralize many bacteria
The protein, known as intelectin-2, also helps to strengthen the mucus barrier lining the digestive tract.
MIT scientists investigate memorization risk in the age of clinical AI
New research demonstrates how AI models can be tested to ensure they don’t cause harm by revealing anonymized patient health data.
Study: High-fat diets make liver cells more likely to become cancerous
New research suggests liver cells exposed to too much fat revert to an immature state that is more susceptible to cancer-causing mutations.
Anything-goes “anyons” may be at the root of surprising quantum experiments
MIT physicists say these quasiparticles may explain how superconductivity and magnetism can coexist in certain materials.
Guided learning lets “untrainable” neural networks realize their potential
CSAIL researchers find even “untrainable” neural nets can learn effectively when guided by another network’s built-in biases using their guidance method.
What makes a good proton conductor?
MIT researchers found a way to predict how efficiently materials can transport protons in clean energy devices and other advanced technologies.
Enabling small language models to solve complex reasoning tasks
The “self-steering” DisCIPL system directs small models to work together on tasks with constraints, like itinerary planning and budgeting.
New method improves the reliability of statistical estimations
The technique can help scientists in economics, public health, and other fields understand whether to trust the results of their experiments.
When it comes to language, context matters
MIT researchers identified three cognitive skills that we use to infer what someone really means.