Google’s Gemini AI Model Demo: Not as Impressive as It Seems
Google recently unveiled Gemini, a new AI model that outperformed OpenAI’s GPT-4 in intelligence benchmarks. A demo video showcasing Gemini’s capabilities, such as tracking a ball in a cup trick and identifying objects, went viral on social media. However, experts soon discovered that the video was edited to enhance Gemini’s performance.
How Google Edited the Gemini Demo
Google confirmed that the demo video was not recorded in real-time. The audio was added later, and the prompts were shortened for brevity. Despite the editing, Google emphasized that the user prompts and outputs in the video were real, aiming to inspire developers with Gemini’s potential.
Despite the backlash over the edited demo video, Gemini has shown promising reasoning skills in extracting scientific data and assisting with academic tasks. Whether users find Gemini’s capabilities impressive or not remains to be seen.


