Caught Red Handed: Google Admits Staging Gemini AI Demo Video

Google recently acknowledged that certain elements of its demonstration video for the Gemini AI, a competitor to OpenAI’s new GPT-4, were staged, raising questions about the actual abilities of the internet giant’s latest artificial intelligence system.

The Verge reports that Google introduced Gemini, its latest and most powerful AI suite, amidst high expectations and fanfare. However, the tech giant found itself in the center of a controversy when it confessed to editing a demonstration video of the AI system, which it admitted shortly after op-ed published by Bloomberg raised suspicions about the video’s authenticity.

The disputed video, presented as a hands-on demonstration, showcased Gemini’s multimodal capabilities, including spoken conversational prompts and image recognition. It depicted the AI recognizing images and responding to them in real-time, a feat that, if genuine, would have marked a significant advancement in AI technology.

However, upon closer inspection and subsequent admission from Google, it was revealed that the demo did not occur in real-time with spoken prompts. Instead, it used still image frames and pre-written text prompts. This has led to criticism from many, including tech columnists like Bloomberg’s Parmy Olson, who stated: “That’s quite different from what Google seemed to be suggesting: that a person could have a smooth voice conversation with Gemini as it watched and responded in real-time to the world around it.”

Google’s previous AI tools have also faced controversy, like the Duplex AI voice assistant’s demo, which many have also called into question due to the AI’s seemingly perfect execution of complicated tasks such as ringing restaurants and making reservations on autopilot. Such incidents have raised questions about the authenticity and reliability of AI demonstrations, especially when edited or pre-recorded.

Despite the controversy, Google maintains that the user prompts and outputs in the Gemini video were real, albeit shortened for brevity. Oriol Vinyals, Vice President of Research and Deep Learning Lead at Google, defended the video, stating it was created to inspire developers and illustrate potential user experiences with Gemini. However, critics argue that true inspiration and confidence in AI technology comes from unedited, real-world demonstrations and public beta testing, rather than carefully crafted promotional videos.

Breitbart News will continue to report on Google’s AI fumbles.

Lucas Nolan is a reporter for Breitbart News covering issues of free speech and online censorship.