Google under fire over edited Gemini AI demo video
Last week Google released Gemini AI, the company’s highly anticipated answer to OpenAI’s ChatGPT. Google called its model “largest and most capable AI model” seemingly suggesting that it’s better than any other AI model that has been developed.
And for the most part, the news was well received. On Thursday, a day after the announcement, Google shares had risen by more than 5%.
However, this didn’t stop Google from being called out over what some users are calling a misleading demo video.
In a 6-minute video released to media outlets and the public, Google demonstrates what is supposedly a conversation between the user and a Gemini-powered chatbot while also showcasing Gemini’s ability to recognize and distinguish between visual pictures and physical objects.
In the video, the chatbot impressively describes the user's actions from placing a piece of paper on the table, to creating an abstract image that turns out to be a duck swimming in water.
The user then proceeds to play games with the chatbot including “Find the object under the cup.”
For the entire clip, the chatbot demonstrates a high level of intelligence and its capabilities are amazing. However, a lot of that comes into question when you realize that the video is not real-time. Meaning there were behind-the-scenes prompts used that we the viewers have no idea about. Prompts that were more specific and detailed than the dialogue in the video.
The video also seems to suggest that a person could have a smooth voice conversation with Gemini as it watched and responded in real-time to the world around it which is not accurate.
As revealed by a post from The Information Google researchers didn’t prompt Gemini with normal speech and video visuals as the video suggests. Instead, they seem to have used a carefully crafted combination of images and text.
The company’s description on YouTube includes a short line that says, “For the purposes of this demo, latency has been reduced, and Gemini outputs have been shortened for brevity.” But, the disclaimer is ambiguous and missing in the video itself.
In its defense, Google has said that the video is a demo of what’s possible with Gemini AI and they’re looking forward to seeing how people put it into use.
“The video is an illustrative depiction of the possibilities of interacting with Gemini, based on real multimodal prompts and outputs from testing. We look forward to seeing what people create when access to Gemini Pro opens on December 13,” a company rep said.