Instead of the AI responding to voice and video, it was fed text prompts and still images.
Continue reading…