Google faces backlash over edited demo of its new AI, Gemini

The controversy emerged when an op-ed from Bloomberg accused Google of misrepresenting the performance of Gemini in a demo video.

The six-minute video, presented during Google's announcement, demonstrated Gemini's multimodal capabilities, combining spoken conversational prompts with real-time image recognition.

However, Bloomberg columnist Parmy Olson claimed that the impressive display may not have been as it seemed. The video, according to Olson, used still image frames from raw footage with written text prompts, rather than actual spoken prompts in real-time.

Google's disclaimer in the video description, stating, "For the purposes of this demo, latency has been reduced, and Gemini outputs have been shortened for brevity," further fueled the controversy.

In response to the accusations, Oriol Vinyals, vice president of research and deep learning lead at Google's DeepMind and co-lead for Gemini, defended the demo. Vinyals explained that all user prompts and outputs in the video are real but shortened for brevity. He emphasized that the video was created to inspire developers and showcase the potential user experiences with Gemini.

The video illustrates what the multimode user experiences built with Gemini could look like. We made it to inspire developers.

Despite Google's explanation, many argue that the controversy undermines the tech giant's credibility, especially given its rivalry with OpenAI's GPT and its enormous success this year.