Vision or multimodality is not strictly necessary for an AI model to function effectively