AI · AI today, November 9, 2026
Tools
FLEX helps robots infer preferences beyond a literal instruction
· Aster editorial team
AI-generated imageAn MIT team introduced FLEX, a framework that uses language to turn human corrections into broader preferences for a robot. The process adds a correction, generalizes it with a language model, and masks parts that should not apply in every context. In simulations involving a robotic arm and tasks with coffee, a candle, and a bowling ball, the system used GPT-4o and took about five seconds to process each interaction.
Why it matters
The approach aims to make one correction useful in related situations without having the robot apply the rule indiscriminately. The test was simulation research, not a deployment in homes or factories.
For the technical crowd
FLEX organizes learning into three steps: add, generalize, and mask. The experiments described by MIT used GPT-4o as the language model.
· Source: MIT News
More from AI today
- Google introduces TPU 8tx and 8i Max to broaden its inference optionsGoogle announced TPU 8tx, a confidential-inference system built on the same silicon as TPU 8t and paired with servers using fifth-generation AMD EPYC…
- Meta expands AI age estimation for teen accounts worldwideMeta announced the worldwide expansion of its age-estimation system on Instagram to detect accounts that may belong to teenagers even when they list…
- LG AI Research opens a free API for K-EXAONE 2.5LG AI Research made K-EXAONE 2.5 available through a free API on its EXAONE platform for developers, startups, companies, and research institutions.…
- WorldParticle uses one model to simulate very different materialsMIT researchers developed WorldParticle, a deep-learning system that represents different kinds of matter as particles and predicts how they move.…
- AI's power demand drives a search for used turbinesUtilities and data-center developers are seeking used or refurbished gas turbines, along with systems adapted from aircraft engines, as new equipment…