Editor's PickQ&A: Enabling more human-like interactions and understanding in multimodal foundation models | MIT News