![gemini](https://wikitechlibrary.com/wp-content/uploads/2023/12/gemini.webp)
Google has admitted to staging parts of its promotional video for the newly launched Gemini chatbot. The tech giant said it edited the speed of Gemini’s responses and added an implied conversational flow that did not occur between the human and AI users.
The original Gemini demo video featured a Google engineer interacting with the chatbot in real time, prompting it with drawings and moving objects.
However, Google revealed that this footage was created “using still image frames from the footage, and prompting via text,” rather than having Gemini respond to visual changes on camera.
While defending the edits as minor, Google acknowledged the misleading nature of the demo video, which lacked any disclaimer about the staged interactions with Gemini.
The edits call into question what Gemini’s real-time capabilities are at this stage, versus what was depicted.
Gemini co-lead Oriol Vinyals stated on Twitter that “all the user prompts and outputs in the video are real,” and that the video’s purpose was “to inspire developers.”
Nonetheless, expectations may have been set too high by the polished demo. Google will need to showcase Gemini’s progress transparently to rebuild trust around the chatbot’s evolving conversational abilities.