Google has acknowledged that the viral video highlighting the capabilities of its artificial intelligence (AI) model, Gemini, was edited to present a more impressive demonstration.
The video, known as the Gemini demo and with 1.6 million views on YouTube, initially showcased real-time interactions of the AI with spoken-word prompts and video.
However, Google clarified in a blog post that the responses were accelerated for the demo’s sake, and the AI wasn’t responding to voice or video but to text prompts using still image frames.
In response to inquiries, a Google spokesperson stated: “Our Hands-on with Gemini demo video shows real prompts and outputs from Gemini. We made it to showcase the range of Gemini’s capabilities and to inspire developers.”
The video, which included scenarios like identifying objects and participating in a game idea generation, was created by capturing footage from the video to test Gemini’s capabilities on various challenges.
While sequences were shortened, still images were used, and the voiceover remained from the written prompts fed into Gemini.
Despite the edited nature of the demo, Google’s AI model remains impressive, showcasing capabilities comparable to OpenAI’s GPT-4.
The release of the video coincided with a period of turmoil in the AI space following Sam Altman’s CEO drama at OpenAI.
The comparison between Google and OpenAI’s advancements in AI remains uncertain, with both companies pushing the boundaries of AI technology.