Archives des National Science Foundation (NSF)

NSF renews support for MIT-led AI and physics institute, expanding a new model for discovery
22 Juin
Teaching AI agents to ask better questions by playing “Battleship”
22 Juin
Solving the “Whac-a-mole dilemma”: A smarter way to debias AI vision models
22 Juin
MIT scientists build the world’s largest collection of Olympiad-level math problems, and open it to everyone
22 Juin
Helping data centers deliver higher performance with less hardware
22 Juin
MIT researchers use AI to uncover atomic defects in materials
22 Juin
Augmenting citizen science with computer vision for fish monitoring
22 Juin
Generative AI improves a wireless vision system that sees through obstructions
22 Juin
3 Questions: On the future of AI and the mathematical and physical sciences
22 Juin
New method could increase LLM training efficiency
22 Juin
Parking-aware navigation system could prevent frustration and emissions
22 Juin
Study: Platforms that rank the latest LLMs can be unreliable
22 Juin
How generative AI can help scientists synthesize complex materials
22 Juin
MIT scientists investigate memorization risk in the age of clinical AI
22 Juin
Guided learning lets “untrainable” neural networks realize their potential
22 Juin
Enabling small language models to solve complex reasoning tasks
22 Juin
New method improves the reliability of statistical estimations
22 Juin
MIT engineers design an aerial microrobot that can fly as fast as a bumblebee
22 Juin
Researchers discover a shortcoming that makes LLMs less reliable
22 Juin
Teaching robots to map large environments
22 Juin
3 Questions: How AI is helping us monitor and support vulnerable ecosystems
22 Juin
Improving the workplace of the future
22 Juin
New tool makes generative AI models more likely to create breakthrough materials
22 Juin
A new generative AI approach to predicting chemical reactions
22 Juin
MIT tool visualizes and edits “physically impossible” objects
22 Juin
New algorithms enable efficient machine learning with symmetric data
22 Juin
The unique, mathematical shortcuts language models use to predict dynamic scenarios
22 Juin
Can AI really code? Study maps the roadblocks to autonomous software engineering
22 Juin
Study could lead to LLMs that are better at complex reasoning
22 Juin
Robotic probe quickly measures key properties of new materials
22 Juin
Unpacking the bias of large language models
22 Juin
Photonic processor could streamline 6G wireless signal processing
22 Juin
With AI, researchers predict the location of virtually any protein within a human cell
22 Juin
New model predicts a chemical reaction’s point of no return
22 Juin
“Periodic table of machine learning” could fuel AI discovery
22 Juin
Training LLMs to self-detoxify their language
22 Juin
Could LLMs help design our next medicines and materials?
22 Juin

NSF renews support for MIT-led AI and physics institute, expanding a new model for discovery

Teaching AI agents to ask better questions by playing “Battleship”

Solving the “Whac-a-mole dilemma”: A smarter way to debias AI vision models

MIT scientists build the world’s largest collection of Olympiad-level math problems, and open it to everyone

Helping data centers deliver higher performance with less hardware

MIT researchers use AI to uncover atomic defects in materials

Augmenting citizen science with computer vision for fish monitoring

Generative AI improves a wireless vision system that sees through obstructions

3 Questions: On the future of AI and the mathematical and physical sciences

New method could increase LLM training efficiency

Parking-aware navigation system could prevent frustration and emissions

Study: Platforms that rank the latest LLMs can be unreliable

How generative AI can help scientists synthesize complex materials

MIT scientists investigate memorization risk in the age of clinical AI

Guided learning lets “untrainable” neural networks realize their potential

Enabling small language models to solve complex reasoning tasks

New method improves the reliability of statistical estimations

MIT engineers design an aerial microrobot that can fly as fast as a bumblebee

Researchers discover a shortcoming that makes LLMs less reliable

Teaching robots to map large environments

3 Questions: How AI is helping us monitor and support vulnerable ecosystems

Improving the workplace of the future

New tool makes generative AI models more likely to create breakthrough materials

A new generative AI approach to predicting chemical reactions

MIT tool visualizes and edits “physically impossible” objects

New algorithms enable efficient machine learning with symmetric data

The unique, mathematical shortcuts language models use to predict dynamic scenarios

Can AI really code? Study maps the roadblocks to autonomous software engineering

Study could lead to LLMs that are better at complex reasoning

Robotic probe quickly measures key properties of new materials

Unpacking the bias of large language models

Photonic processor could streamline 6G wireless signal processing

With AI, researchers predict the location of virtually any protein within a human cell

New model predicts a chemical reaction’s point of no return

“Periodic table of machine learning” could fuel AI discovery

Training LLMs to self-detoxify their language

Could LLMs help design our next medicines and materials?