Google is under scrutiny for its demonstration video for its newly launched artificial intelligence model, Gemini.
On Wednesday, just weeks before the end of the year, Google unveiled what it believes is its largest and most powerful AI model, Gemini, and presented a demonstration video to the media and public.
The six-minute video includes spoken conversations between the user and a Gemini-powered chatbot, and also showcases Gemini’s ability to recognize visual images and physical objects and tell the difference. Some of the skills were impressive, such as Gemini’s ability to describe drawings of a duck out loud and describe the drawing of a duck versus a rubber duck, to name a few.
The company’s description on YouTube includes a short line that says, “For the purposes of this demo, latency has been reduced and Gemini output has been shortened for brevity.” However, this disclaimer is not made in the video itself.
After the launch, the company later confirmed to Bloomberg that the demo was not conducted in real time, but instead used still images and fed text prompts, which Gemini responded to, as The Information previously highlighted. The author noted that this was “very different” than what Google seemed to be suggesting: “That a person could have a smooth voice conversation with Gemini as it observed and responded to the world around them in real time.”
After multiple requests for comment, the company told CNBC in a statement on Friday: “The video is a vivid representation of the possibilities of interacting with Gemini, based on real multimodal prompts and test results. We look forward to seeing what people create and when.” Access to Gemini Pro begins December 13th.”
Although demos are often edited, Gemini’s subsequent results represent déjà vu for the search giant.
Earlier this year, Google was criticized by the public and Wall Street for what its own employees called a “rushed, botched” demonstration of its AI chatbots. The demonstration took place the same week that Microsoft planned to unveil its Bing integration with ChatGPT.
Earlier this month, The Information reported that Google scrapped plans for a series of in-person events to launch Gemini, ultimately opting for a virtual launch.
Google faces stiff competition from Microsoft-backed OpenAI’s GPT-4, which has been the most advanced and successful model to date. Google released a white paper this week claiming that Gemini’s top-performing “Ultra” model outperformed GPT-4 in several benchmarks, albeit incrementally.
Don’t miss these stories from CNBC PRO:
Source : www.cnbc.com