Microsoft is introducing an experimental feature called “Portraits” within its Copilot Labs, designed to provide a more engaging experience when interacting with its AI assistant. This feature is currently available in the US, UK, and Canada and includes 40 stylized human avatars that are intended to respond with natural expressions during real-time voice conversations.
Mustafa Suleyman, CEO of Microsoft AI, announced the new feature on X, stating that it was developed in response to feedback from some users who expressed a preference for conversing with a visual representation rather than solely through text. The Portraits feature allows users to choose an avatar and pair it with a voice, aiming to create a more natural interaction compared to text-based communication.
This initiative is part of Microsoft’s ongoing efforts to make Copilot a more relatable entity, following their previous announcement of “Copilot Appearances” in July. The main distinction between these two features lies in the visual representations, with Portraits focusing on human-like appearances instead of cartoonish designs.
The underlying technology of this feature leverages the VASA-1 system, a sophisticated AI developed by Microsoft Research. This technology facilitates real-time, visual AI conversations, generating believable facial expressions, head movements, and lip-syncing capabilities using just a single image, thus avoiding the need for complex 3D modeling.
Additionally, Microsoft emphasizes that each avatar is stylized rather than photorealistic, aiming to maintain visual variety while ensuring the trusted intelligence and safety features associated with Copilot. The feature is currently accessible to a limited user base aged 18 and older, with certain restrictions in place, including time limits and notifications indicating the interaction is with AI technology.
Source: https://www.theverge.com/news/788741/microsoft-copilot-labs-ai-portraits-experiment-human-avatar