In the increasingly fierce competition in the generative AI market,GoogleRecently launched its large language model Gemini However,The controversy over the authenticity of the video subsequently attracted widespread attention.
The demo video released by Google shows the multimodal capabilities of the Gemini model, which can cleverly interpret and process information from real-time video and audio. This is a major achievement for Google, especially in the fierce competition with rivals such as OpenAI. However, according to Bloomberg, the demo video was actuallyMade using still image frames from a video and text prompts, rather than the real-time voice and video processing that seems to be achieved.
Sissie Hsiao, vice president and general manager of Google Assistant and Bard, discussed the controversial demo video at Fortune’s Brainstorm AI conference in San Francisco, highlighting the standards Gemini has achieved as a model and how it will advance the development of Google’s chatbot, Bard.“This video is completely real. All the prompts and model responses are real,” Hsiao said. “We did shorten some parts for brevity, but that information is also explained in the video.”
The demonstration video shows the multimodal capabilities of the new AI model, which recognizes a wavy line, then recognizes the curve of a new line, and finally draws a picture of a duck. Throughout the process, the model continues to recognize each element and provide facts and answers related to the duck in real time.
Hsiao highlighted Gemini’s achievements in a variety of benchmarks, ranging from high school physics to professional legal puzzles and ethical scenarios. According to The Verge, Gemini Ultra beat OpenAI’s GPT-4 in 32 benchmarks, winning 30 tests in total, an achievement worth boasting about, even though Gemini Ultra won’t be released until next year.Currently, Bard uses the less advanced Gemini Pro, which is roughly equivalent to GPT 3.5.
Hsiao said these Gemini models will continue to improve Google Search as well as the Google Bard chatbot, which she said is “the most advanced chatbot on the market today.”Most PopularFree chatbot for .