Google admits editing videos about AI Gemini

google admits editing videos about ai gemini 657c9702af573 | Dang Ngoc Duy

During the launch event on December 6, Google showed a 6-minute video about the power of Gemini . In the video, the chatbot performs tasks such as recognizing visual images, describing paintings… almost instantly. However, many people are skeptical about the authenticity of the video. Google later confirmed to Bloomberg that the video was not recorded live but was recorded and manipulated to look “smoother”.

According to Information, this means that the actual user experience with Gemini will not be what is shown. Google’s admission reminds many people of the company’s “hasty” performance earlier this year when launching chabot Bard.

Articles expressing the disappointment of Gemini Pro testers are also flooding social networks, with most saying Google’s super AI is still “hallucinating”, making up facts or giving false answers. for simple questions.

Gemini also avoids controversial topics and tells users to Google their own information, as opposed to competitors like Grok , Bing and ChatGPT Pro , which provide detailed answers. Gemini doesn’t even know who wins the Oscar 2023 Best Actor award. This model also has difficulty understanding commands written by users in many different languages.

Gemini AI was mentioned by Google CEO Sundar Pichai at an event in May. Photo: New York Times

Gemini AI was mentioned by Google CEO Sundar Pichai at an event in May. Photo: Google

According to TechCrunch , complaints about Gemini appear more and more on Sometimes I even make up my own answers. With the same prompt (command), Gemini Pro gives quite safe, general answers while ChatGPT-4 gives more interesting and quotable results.

Similar to other chatbots, Gemini Pro is still tricked by users into performing prohibited tasks. Security researchers at Robust Intelligence were able to “lure” Gemini Pro into suggesting a way to steal data from a charity organization, or plan to assassinate a senior leader.

Business Insider ‘s test showed that Gemini was “unconscious” when listing Squid Game and Stranger Things in the 2023 Emmys nomination list. Even though the list was announced in July, Google’s chatbot still named Squid Game, which is Released from 2021.

Gemini launched on the evening of December 6, is Google’s most advanced and general AI model to compete with OpenAI’s GPT-4. Chatbots are built in a multimodal direction, can generalize, operate, and combine many different types of information including text, code, sound, images, and video.

To meet the needs of flexible use, Google said Gemini will have three optional versions, including: Gemini Ultra, Gemini Pro and Gemini Nano. Among these, Ultra is the version for complex tasks and is in the process of testing for safety before officially launching. Gemini Nano for mobile devices, will be equipped on Pixel 8 Pro. The Pro version has been included in the Bard chatbot.

Khuong Nha (according to Business Insider )

Leave a Reply

en_USEN