Google has launched Gemini 2.5 Computer Use, an advanced AI model that simulates human-like interactions with software. Unlike traditional approaches relying on structured APIs, this model features visual understanding, allowing it to execute tasks such as filling out forms and scrolling through websites. The Gemini 2.5 model operates through a continuous loop, analyzing screenshots, user requests, and action history to generate UI actions. With capabilities to perform 13 different tasks, it excels in web browser and Android mobile environments, although it’s not yet optimized for desktop controls. Google claims it outperforms competitors like Claude and ChatGPT in several benchmarks, demonstrating enhanced speed and efficiency — with early users reporting performance boosts up to 50%. To address potential AI risks, Google has implemented safety features that prevent high-risk actions. Available in preview via Google AI Studio and Vertex AI, users can explore demos showcasing its functionality in tasks like gaming and web browsing.
Source link
