MIT researchers develop an efficient way to train more reliable AI agents
The technique could make AI systems better at complex tasks that involve variability.
Topic
Laboratory for Information and Decision Systems (LIDS)
Download RSS feed: News Articles / In the Media / Audio
A causal theory for studying the cause-and-effect relationships of genes
By sidestepping the need for costly interventions, a new method could potentially reveal gene regulatory programs, paving the way for targeted treatments.
Despite its impressive output, generative AI doesn’t have a coherent understanding of the world
Researchers show that even the best-performing large language models don’t form a true model of the world and its rules, and can thus fail unexpectedly on similar tasks.
Helping robots zero in on the objects that matter
A new method called Clio enables robots to quickly map a scene and identify the items they need to complete a given set of tasks.
3 Questions: Should we label AI systems like we do prescription drugs?
Researchers argue that in health care settings, “responsible use” labels could ensure AI systems are deployed appropriately.
Study: AI could lead to inconsistent outcomes in home surveillance
Researchers find large language models make inconsistent decisions about whether to call the police when analyzing surveillance videos.
MIT researchers use large language models to flag problems in complex systems
The approach can detect anomalies in data recorded over time, without the need for any training.
Study: When allocating scarce resources with AI, randomization can improve fairness
Introducing structured randomization into decisions based on machine-learning model predictions can address inherent uncertainties while maintaining efficiency.
Large language models don’t behave like people, even though we may expect them to
A new study shows someone’s beliefs about an LLM play a significant role in the model’s performance and are important for how it is deployed.
AI model identifies certain breast tumor stages likely to progress to invasive cancer
The model could help clinicians assess breast cancer stage and ultimately help in reducing overtreatment.
How to assess a general-purpose AI model’s reliability before it’s deployed
A new technique enables users to compare several large models and choose the one that works best for their task.
Machine learning and the microscope
PhD student Xinyi Zhang is developing computational tools for analyzing cells in the age of multimodal data.
When to trust an AI model
More accurate uncertainty estimates could help users decide about how and when to use machine-learning models in the real world.
The tenured engineers of 2024
Twelve faculty members have been granted tenure in six units across MIT’s School of Engineering.
Study reveals why AI models that analyze medical images can be biased
These models, which can predict a patient’s race, gender, and age, seem to use those traits as shortcuts when making medical diagnoses.