Google has unveiled a global update to its Bard chatbot, Gemini, which includes a new "multimodal input" feature. A YouTube video published by Google demonstrates user interaction with the chatbot through voice commands, gestures and images.
The "multimodal input" feature allows users to communicate effectively with the chatbot by providing queries in a variety of formats. The video quickly highlights the high quality of the model's perception and accuracy of responses, which has elicited positive feedback from many users.
However, some members of the community have begun to raise doubts about the authenticity of the information presented. An artificial intelligence expert under the pseudonym "Nelly R Q" raised suspicions of deception on Google's part, supported by Bloomberg Opinion columnist Parmi Olson. Olson suggested that the events in the video may have been pre-recorded rather than happening in real time.
These suspicions were confirmed by a statement from network engineer and journalist Chief Nerd, who suggested that the video had been sped up. Google OpenMind spokesperson Oriol Viñals later confirmed that the video used accelerated prompts to reduce the model's response time. Viñals said that the purpose of the video was to inspire developers and explained the changes in the process of preparing the answer output.
We can't fail to mention that in November 2023, the chatbot was also unveiled by xAI, a company owned by Ilon Musk, which led to a sudden rise in the price of the eponymous token by 1000% in a matter of days.
