Google has been met with criticism regarding its six-minute promotional video of Gemini, which was intended to show the preeminence of its AI technology. Following Gemini's launch, the video aroused attention from many people, calling to mind the rushed and imperfect demonstrations the company conducted earlier this year and which their own staff labeled "botched."
Google is facing scrutiny over its recently released demonstration video for its artificial intelligence model, Gemini. On Wednesday, just before the year's end, the company introduced what it believes is its biggest and most capable AI model and exhibited a demonstration video to the public and media outlets. The six-minute video featured conversations between the user and the Gemini-powered chatbot and illustrated Gemini's capacity to recognize physical objects and pictures. Gemini was able to voice aloud a description of drawings of a duck, plus distinguish between a drawing of a duck and a rubber duck. Google's YouTube description has a line that says, "For the purposes of this demo, latency has been reduced, and Gemini outputs have been shortened for brevity." Nevertheless, the video does not contain this disclaimer. Following the launch, Google confirmed to Bloomberg that it was not a real-time demo, but rather still images and text prompts with artificial responses, as was previously noted by The Information. The author commented that this was "quite different" from what Google seemed to demonstrate: "that a person could have a smooth voice conversation with Gemini as it watched and responded in real-time to the world around it." After a couple of requests for comments, the company attested to CNBC in a statement, "The video is an illustrative depiction of the possibilities of interacting with Gemini, based on real multimodal prompts and outputs from testing. We look forward to seeing what people create when access to Gemini Pro opens on December 13." Although demonstrations are usually edited, the results of Gemini elicited déjà vu for the search giant. Earlier this year, Google and its own employees endured criticism from the public and Wall Street for what they called a "rushed, botched" demonstration of its AI chatbots, around the same time Microsoft planned to show its Bing integration with ChatGPT. In the beginning of this month, The Information reported that the tech giant terminated plans for a series of in-person events to launch Gemini, and eventually chose a virtual launch. Google is in fierce competition with OpenAI's GPT-4, Microsoft-supported, which until now has been the most advanced and productive model. This week Google published a white paper claiming Gemini's "Ultra" model, its most powerful, outperformed GPT-4 in various tests, though marginally.
Commentaires