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.

Adam Zewe | MIT News • mit
June 23, 2025 ~7 min

Unpacking the bias of large language models

In a new study, researchers discover the root cause of a type of bias in LLMs, paving the way for more accurate and reliable AI systems.

Adam Zewe | MIT News • mit
June 17, 2025 ~8 min


Inroads to personalized AI trip planning

A new framework from the MIT-IBM Watson AI Lab supercharges language models, so they can reason over, interactively develop, and verify valid, complex travel agendas.

Lauren Hinkel | MIT-IBM Watson AI Lab • mit
June 10, 2025 ~9 min

AI-enabled control system helps autonomous drones stay on target in uncertain environments

The system automatically learns to adapt to unknown disturbances such as gusting winds.

Adam Zewe | MIT News • mit
June 9, 2025 ~8 min

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

Trained with a joint understanding of protein and cell behavior, the model could help with diagnosing disease and developing new drugs.

Adam Zewe | MIT News • mit
May 15, 2025 ~8 min

Study shows vision-language models can’t handle queries with negation words

Words like “no” and “not” can cause this popular class of AI models to fail unexpectedly in high-stakes settings, such as medical diagnosis.

Adam Zewe | MIT News • mit
May 14, 2025 ~8 min

New tool evaluates progress in reinforcement learning

“IntersectionZoo,” a benchmarking tool, uses a real-world traffic problem to test progress in deep reinforcement learning algorithms.

MIT Laboratory for Information and Decision Systems • mit
May 5, 2025 ~7 min

Designing a new way to optimize complex coordinated systems

Using diagrams to represent interactions in multipart systems can provide a faster way to design software improvements.

MIT Laboratory for Information and Decision Systems • mit
April 24, 2025 ~10 min


A faster way to solve complex planning problems

By eliminating redundant computations, a new data-driven method can streamline processes like scheduling trains, routing delivery drivers, or assigning airline crews.

Adam Zewe | MIT News • mit
April 16, 2025 ~8 min

Researchers teach LLMs to solve complex planning challenges

This new framework leverages a model’s reasoning abilities to create a “smart assistant” that finds the optimal solution to multistep problems.

Adam Zewe | MIT News • mit
April 2, 2025 ~7 min

/

12