A video showcasing Google’s artificial intelligence (AI) model, known as Gemini, has come under scrutiny for misleading representations. The demo video, which garnered 1.6 million views on YouTube, demonstrates a back-and-forth interaction where the AI responds in real-time to spoken-word prompts and video. However, Google admitted that it had sped up responses for the demo and revealed that the AI was not responding to voice or video but was prompted using still image frames and text. The company clarified that the video aimed to showcase the range of Gemini’s capabilities.
In the demo, the AI correctly identifies objects and answers questions posed by the demonstrator. Still, the method used involves showing the AI a still image of an object and providing a text prompt, deviating from the initial impression of real-time interaction with video content. Google defended the video, stating it was created to test Gemini’s capabilities on various challenges and to inspire developers.
The demo raises questions about the transparency and accuracy of AI demonstrations and highlights the need for clear communication about the capabilities of AI models. While Google’s AI remains impressive, the use of still images and text-based prompts aligns it with models like OpenAI’s GPT-4. The video release comes amid ongoing developments and competition in the AI space, including OpenAI’s work on the next version of its AI model.