A new way to edit or generate images
MIT researchers found that special kinds of neural networks, called encoders or “tokenizers,” can do much more than previously realized.
Computer science and technology
The unique, mathematical shortcuts language models use to predict dynamic scenarios
Language models follow changing situations using clever arithmetic, instead of sequential tracking. By controlling when these approaches are used, engineers could improve the systems’ capabilities.
This “smart coach” helps LLMs switch between text and code
The CodeSteer system could boost large language models’ accuracy when solving complex problems, such as scheduling shipments in a supply chain.
Can AI really code? Study maps the roadblocks to autonomous software engineering
A team of researchers has mapped the challenges of AI in software development, and outlined a research agenda to move the field forward.
How to more efficiently study complex treatment interactions
A new approach for testing multiple treatment combinations at once could help scientists develop drugs for cancer or genetic disorders.
AI shapes autonomous underwater “gliders”
An AI pipeline developed by CSAIL researchers enables unique hydrodynamic designs for bodyboard-sized vehicles that glide underwater and could help scientists gather marine data.
Study could lead to LLMs that are better at complex reasoning
Researchers developed a way to make large language models more adaptable to challenging tasks like strategic planning or process optimization.
Robotic probe quickly measures key properties of new materials
Developed to analyze new semiconductors, the system could streamline the development of more powerful solar panels.
Using generative AI to help robots jump higher and land safely
MIT CSAIL researchers combined GenAI and a physics simulation engine to refine robot designs. The result: a machine that out-jumped a robot designed by humans.
LLMs factor in unrelated information when recommending medical treatments
Researchers find nonclinical information in patient messages — like typos, extra white space, and colorful language — reduces the accuracy of an AI model.