“Erick Friis is personally excited about the emergence of models capable of mixed-modality output (text, images, audio), viewing them as a vision for the future despite their current lack of reliability.”