MIT uses two LLMs together to help robots parse vague instructions

MIT CSAIL pairs one language model to clarify fuzzy human instructions with another that filters out irrelevant details, helping robots do household and factory chores.