Google is facing intense scrutiny over a demonstration video of its newly launched artificial intelligence model Gemini.
On Wednesday, just a few weeks before the end of the year, Google unveiled what it considers to be its largest and most capable AI model, Gemini, and released a demo video to the media and public.
The 6-minute video includes a conversation between a user and a Gemini-powered chatbot, and also demonstrates Gemini’s ability to recognize and tell the difference between visual images and physical objects. . Some of the abilities were impressive, such as Gemini’s ability to describe aloud the description of a picture of a duck, and describe a picture of a duck and a rubber duck.
In the company’s description, YouTube It includes a short line that says, “For the purposes of this demo, Gemini’s output has been shortened for latency and brevity.” However, the video itself contains no such disclaimer.
After launching, the company later confirmed According to Bloomberg, the demo was not done in real time, but instead used static images and text prompts for Gemini to respond to, as before. pointed It came from The Information. The author states that this is “completely different” than what Google seems to suggest, adding that “a person can communicate smoothly with Gemini while the Gemini monitors and responds to the world around them in real time.” It’s about being able to have a conversation.”
After multiple requests for comment, the company said in a statement to CNBC on Friday: “This video illustratively depicts the potential for interaction with Gemini, based on actual multimodal prompts and output from testing. We look forward to seeing what people create and when.” Access to Gemini Pro will begin on his December 13th. ”
Although demos are often redacted, Gemini’s subsequent discovery brings a sense of déjà vu to the search giant.
It comes in the same week that Google faced public and Wall Street criticism earlier this year after its own employees staged a “rushed and botched” demonstration of an AI chatbot. microsoft I was going to introduce the Bing and ChatGPT integration.
Earlier this month, The Information reported that Google: canceled plans We planned a series of in-person events to launch Gemini, eventually settling on a virtual launch.
Google is in fierce competition with Microsoft-backed OpenAI’s GPT-4, the most advanced and successful model to date. Google released a white paper this week claiming that Gemini’s most powerful model, Ultra, outperforms GPT-4 against several benchmarks, albeit incrementally.
Don’t miss the next story from CNBC PRO.