We let Chrome’s Auto Browse agent surf the web for us—here’s what happened (arstechnica.com)

Google has introduced its new Auto Browse agent as part of Chrome, currently in preview for AI Pro and AI Ultra subscribers. This AI agent aims to enhance user productivity by automating web tasks, a step that reflects an ongoing shift in the AI landscape from basic chatbots to more capable agents that handle diverse online activities. Testing revealed that while Auto Browse can navigate web games and perform specific tasks, it still struggles with some functionalities, such as using arrow keys and interpreting game mechanics like a human would. The implications of this rollout are significant for the AI/ML community, demonstrating both the potential for AI to streamline online tasks and the challenges that remain in developing intuitive, human-like behavior in machine agents. During testing, Auto Browse successfully engaged with a web game but exhibited limitations, such as taking too long to process moves and misinterpreting task parameters. These challenges highlight the ongoing need for refinement in AI agents, as they become increasingly integrated into everyday digital experiences. Overall, while Auto Browse shows promise, it underscores the work needed for AI to truly deliver on its capabilities.