The tech giant clarified that the demo was created to test Gemini's capabilities on a wide range of challenges. It also admitted that the AI did not invent a game as shown in the video, but was given specific instructions. Despite the revelations, Google's AI model is still considered impressive, with capabilities similar to OpenAI's GPT-4.
Key takeaways:
- Google's Gemini demo video, which showcases the capabilities of its AI model, has been revealed to not be as it seems, with the AI not responding to voice or video at all.
- The AI was actually prompted using still image frames from the footage and text, rather than real-time spoken-word prompts and video as the video suggests.
- Despite the misleading representation, Google's AI model is still impressive, with capabilities similar to that of OpenAI's GPT-4.
- The video was released amidst a period of chaos in the AI space, following Sam Altman's firing and rehiring as CEO of OpenAI, and it is unclear which of the two AI models is more advanced.