AI Robotics & Autonomy
FLEX lets robots infer unstated human preferences
The AI system turns conversational corrections into broader rules for robot motion planning.
A robot arm carries a cup across a table while avoiding a laptop and phone.
Photo: Kronos News
Researchers from MIT and New York University developed FLEX, a system that uses a language model to help robots infer preferences beyond a person’s literal instructions. For example, after being told not to spill on a laptop, a robot could also avoid a nearby phone. [1]
MIT says FLEX was tested in simulations involving tasks such as serving coffee and moving objects. The system generalized corrections to categories such as electronics, flammable objects, and fragile items, then updated the robot’s planned path. The report describes simulated tests and does not establish how well FLEX works in real-world settings. [1]
Editorial notes
Transparency note
AI assisted drafting. Human edited and reviewed.
- AI assisted
- Yes
- Human review
- Yes
- Last updated
Risk assessment
The supplied MIT report describes simulated evaluations but does not verify the brief’s claim of a real-robot study.
Sources
Related stories
View all
Topics
About the author
Kronos News Desk covers ai robotics & autonomy and editorial analysis for Kronos News.