This proposal introduces a concept for enhancing ChatGPT by incorporating a user-controlled visual channel that allows for bidirectional communication. The idea is to integrate nonverbal cues, such as facial expressions and gestures, into interactions with AI, alongside traditional text and voice interfaces. This approach aims to improve the clarity of conversations and reduce misunderstandings by providing additional context, with privacy always being prioritized. The initiative is experimental, aiming to test whether visual interaction can significantly enhance the effectiveness and interpretability of human-AI communication.