I Created a Duplicate of Myself Using Gemini’s AI Avatar Tool, and the Outcome Was Eerily Accurate.

It’s a lovely, warm afternoon at Dolores Park in San Francisco, and I’m performing a birthday song for a prehistoric dinosaur. Miraculously, a cupcake with a pink candle appears in my empty hand as I complete my serenade. When I blow out the flame, a serene expression of satisfaction spreads across the CGI-like creature.

While the individual in this AI-generated video resembles me in both looks and voice, the clip was actually created using one of the latest features in Google’s Gemini app: avatars. These digital replicas are akin to the core functionalities of OpenAI’s now-retired Sora app. They serve as a digital version of yourself that can be featured in AI videos. Avatars are powered by the company’s new Omni video model, and this feature is accessible only to subscribers.

I subscribe to Google’s AI Pro plan for $20 a month and quickly hit Gemini’s usage limits, which reset every 5 hours. I simply posed a few questions and created two 10-second clips with my avatar before I received a message to wait until later.

My initial experiences with what Omni can do with my likeness included scenes of me singing to a dinosaur in San Francisco and surfing beneath the Golden Gate Bridge. I felt both amazed and unsettled. The content was cringeworthy, featuring some awkward moments and bizarre outfits, but it was undeniably me in the video. I zoomed in on its face to closely observe the mouth movements. The teeth were somewhat off, but aside from that, it was Reece—down to the chin fat.

Unlike OpenAI, which allowed users to choose whether they wanted their likeness used in AI videos, Google restricts adult users to creating videos with their own avatars only.

Setting up my avatar through the Gemini app took about five minutes. The process required me to be in a well-lit area with my phone’s camera focused on my face while reading a series of two-digit numbers. After that, I turned slowly to the right and then to the left, and that was it. Reece 2.0 was born and ready to take the spotlight as my deepfake star. (Keep an eye on your outfit during this setup, as what you wear may show up in the AI creations, but more on that later.)

Let’s analyze the birthday clip frame by frame to fully explore my feelings. Full prompt: Generate a video of me singing the happy birthday song to an aging dinosaur at the top of the hill at Dolores Park.

The first second opens with a classic pause because even AI Reece retains some ingrained habits. What stands out initially is the photorealistic backdrop. Instead of placing my avatar on a random oversized hill at a park, Google’s AI video features a setting strikingly similar to the actual location. From the palm tree-lined sidewalks to the towering Salesforce in the background, it’s instantly clear which park is portrayed here, despite the fact that the output isn’t flawless. It’s logical that a company skilled in mapping the globe could achieve this.

As AI me began singing, with a less pitchy baritone than I can actually manage, the first few notes felt natural. I moved my hands up and down on the beat, resembling a small conductor. However, I stumbled on the word “to,” and Gemini shifted to a wider angle as the real chaos ensued. A vanilla cupcake suddenly appeared, and I puffed out a cloud of smoke to extinguish the celebration candle. (Honestly, how inconsiderate of AI Reece. It’s not your special day.)