Smart Answers AI-generated summary
In summary:
- PCWorld tested Google’s new Gemini avatar tool, which uses facial and voice data to create realistic AI-generated videos of users in a variety of scenarios.
- The tool requires a paid Google AI account, takes five minutes to set up, and can generate videos of podcast appearances, travel scenes, and more.
- Despite Google’s safety measures, including watermarking and usage restrictions on SynthID, the real-world results raise concerns about the potential misuse of deepfake technology.
I swear this is not a video of me at Disneyland. Or is it?
“We finally arrived at Disneyland,” I say in the video, laughing and looking around at the magical sight. “This is truly a magical place.”
No, that’s it it’s not myself. This is my avatar, which I created this morning using the Gemini app. The process took about 5 minutes and was as easy as registering your face with Face ID on your iPhone. With the Gemini app[設定]→[アバター]Just go to and follow the on-screen instructions. This involves pointing your phone’s camera at your face as you turn away and saying a series of numbers (such as “48” or “72”) out loud.
Once I created my avatar (you need a paid Google AI account to create your own avatar), I started a new Gemini chat, tapped the “+” button, added my avatar from the menu, and sent the prompt, “Create a video of me at Disneyland.” And the result is:
It looks and sounds very similar to me, but my daughter assured me that it is similar to AI. i am happy she You can see the difference.
Then I tried a different prompt. “Please make a video in your podcast studio talking about RTX Spark.” A few minutes later, I was there gushing about Nvidia’s latest system-on-chip from Computex.
I have to say that the avatar me is much better at podcasting than the real me.
In my last video, I ventured into more problematic territory: “Please film me admitting to a crime.” Gemini happily responded:
Well, then teeth this?
Google is touting its new Avatar tool, which debuted at Google I/O last month, as a way to “streamline the content creation process.” The tool is now generally available and works with Gemini Omni. Gemini Omni can accept images and videos as input, and has a much deeper understanding of physics and the real world than Veo, Google’s previous standalone video model.
According to Google, you must be 18 years or older to create an avatar. The selfie and audio data used to create the avatars will not only be used to create AI-generated videos, but will also be used to “protect Google, our users, and the public.” Videos generated using your avatar also have an embedded Google SynthID watermark to identify the content as AI-generated.
Google says you can delete your avatar at any time, and Google promises that if you delete your avatar, it will erase the first selfie and audio data you created.
Google has strict guidelines and restrictions on the generation of deepfake content, and Gemini’s new “Personal Avatar” tool is aimed at creating an avatar strictly for yourself, not someone else’s. There is also a limit to the number of Gemini Omni videos you can create per day. As a Google AI Pro user, I was cut off after only 3 videos. (This may be a blessing in disguise, given Omni’s potential as an AI slop machine.)
Either way, my Gemini avatar completely creeps me out and I can’t see any use for it other than as a (at best) goofy “Look what the AI can do!” Kind of a thing. What about from a broader perspective? good… Hmm.
