Google Gemini Testing 3D AI Avatar Feature

Google is testing a new feature in its Gemini app that allows users to create 3D digital avatars of themselves. Code found in the latest Android beta, version 17.4.66, reveals a tool dubbed "Characters" that prompts users to record a "Likeness" via a selfie video to generate their digital twin [1].

The feature appears in the overflow menu of the Gemini launcher, though the ability to record a fresh video is not yet functional for users [1]. While Google has not confirmed the exact utility of these avatars, the company is positioning them as a potential substitute for users who prefer not to appear on camera during video calls [1]. This approach mirrors the "Persona" feature introduced by Apple for the Vision Pro headset and the "Likeness" tool found on the Galaxy XR [1].

The move signals a broader industry push to integrate AI-powered digital twins into daily workflows. IgniteTech, for instance, has already deployed a service called MyPersona that creates digital clones of subject matter experts for corporate use [2]. By training these clones on specific datasets—such as human resources or technical support policies—businesses aim to reduce the administrative burden on staff [2].

These digital twins are designed to mimic a user’s voice, facial expressions, and speaking cadence to provide a more human-like interaction than standard text-based chatbots [2]. However, the technology remains limited. Users have noted that while these clones can accurately regurgitate information from a provided knowledge base, they often lack the nuance, personality, and humor of their human counterparts [2].

There is also a persistent risk of AI "hallucinations" if a clone is fed inaccurate information, and the rapid pace of technological change means these digital versions require constant updates to remain relevant [2]. Furthermore, the rise of hyper-realistic avatars raises questions about the dangers of further anthropomorphizing AI, which remains a non-human tool regardless of its visual fidelity [2].

As Google moves to bring these digital substitutes to the mass market, the primary challenge will be balancing the convenience of automated representation with the limitations of current AI models. Whether these avatars will become a standard tool for digital communication or remain a niche experiment depends on how effectively they can replicate human nuance without crossing into the uncanny.