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.
Research
A new way to increase the capabilities of large language models
MIT-IBM Watson AI Lab researchers developed an expressive architecture that provides better state tracking and sequential reasoning in LLMs over long texts.
A “scientific sandbox” lets researchers explore the evolution of vision systems
The AI-powered tool could inform the design of better sensors and cameras for robots or autonomous vehicles.
“Robot, make me a chair”
An AI-driven system lets users design and build simple, multicomponent objects by describing them with words.
Deep-learning model predicts how fruit flies form, cell by cell
The approach could apply to more complex tissues and organs, helping researchers to identify early signs of disease.
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.
New materials could boost the energy efficiency of microelectronics
By stacking multiple active components based on new materials on the back end of a computer chip, this new approach reduces the amount of energy wasted during computation.
MIT researchers “speak objects into existence” using AI and robotics
The speech-to-reality system combines 3D generative AI and robotic assembly to create objects on demand.
A smarter way for large language models to think about hard problems
This new technique enables LLMs to dynamically adjust the amount of computation they use for reasoning, based on the difficulty of the question.