I created myself with Gemini’s AI Avatar Tool. The result made me angry

It’s beautiful, It’s a beautiful afternoon in Dolores Park in San Francisco, and I’m singing the birthday song of an ancient dinosaur. A cake with a pink candle appears in my empty hand as I finish my serenade. When I turn off the flame, a gentle look of content washes over the CGI-esque creature.

While the person in the AI ​​video looks and sounds like me, the cartoon was actually made using one of the new features available in Google’s Gemini app: avatars. These digital simulations are similar to previous OpenAI simulations Sister app. It’s a digital version of you that can be embedded into AI videos. Avatars are supported by the new Omni video format, and the feature is only available to subscribers.

I pay $20 a month Google’s AI Pro system and quickly remove the Gemini limit, which resets every 5 hours. I just asked a few questions and made two 10-second photos with my avatar, before I was told to wait until later.

My first two photos that Omni could do with my image were of me playing dino in San Francisco and swimming under the Golden Gate Bridge. I was immediately impressed and nervous. The content was depressing, with flashy moments and ridiculous costumes, but the person in the video was me. I used my fingers to look at his face and watch his mouth move. His teeth were longer, but otherwise he was Reece, down to the fat on his chin.

In contrast OpenAIwhich previously allowed users to choose whether they wanted others to create AI videos using their features, Google only allows adult users to create videos with their avatar.

It took me about five minutes set my avatar through the Gemini program. The method involved sitting in a well-lit room with my phone’s camera pointed at my face and reading a list of two numbers. Then I slowly looked to the right and turned my head to the left, and everything was over. Reece 2.0 was born and he is ready to be my star of the deep. (Have fun with what you’re wearing at this point, because your fit will show up in AI generations, but more on that later.)

Let’s break down the inherent picture and framework to get my thoughts out here. General information: Make a video of me singing happy birthday to an old dinosaur on top of a hill in Dolores Park.

The first two begin with a rest for a thousand years because even AI Reece has some fixed habits. The first thing that impresses is the photorealistic setting. Instead of placing my avatar on a high mountain in the park, the background of the Google AI video is much more similar to the location. From the palm-lined streets to Salesforce looming in the distance, it’s clear which park is on display here, even if the results aren’t perfect. It makes sense that a company known for creating maps of the world could produce this.

When the AI ​​started singing, with a baritone that was less than I could handle, the first few bars seemed natural. I moved my hands up and down to the beat, like a mini conductor. Then, I stutter on the word “to,” and Gemini cuts in as the real confusion begins. Vanilla cake appears randomly, and I blow out a cloud of smoke to blow out the festive candle. (Honestly, how rude AI Reece is. It’s not your special day.)