Google's Chrome browser recently rolled out its Auto Browse agent, a new feature for AI Pro and AI Ultra subscribers. This agent is designed to surf the web and complete tasks on a user's behalf. Given Chrome's dominance, this agent has a potentially enormous reach. The Ars Technica team tested the agent's capabilities with tasks like playing a web game and creating a music playlist.

In the web game test, the agent managed to play 2048, though it struggled initially due to the lack of arrow key support. It ultimately achieved a score and was rated 8/10. The agent's performance was not as good as its competitors, but it required less user input to complete the task. The playlist creation test highlighted some limitations. Specifically, the agent had trouble with YouTube Music and was unable to monitor a live broadcast for an extended period.

The inability to use arrow keys for some tasks and its struggle with YouTube Music raise questions about Auto Browse's current capabilities. The agent could successfully work with Spotify. While the agent shows promise, it's clear that there's still room for improvement in terms of functionality and platform compatibility. Google is likely to continue refining the agent. Ultimately, the success of AI agents hinges on their ability to handle real-world tasks.

Auto Browse's limitations suggest that these AI tools are still in their early stages. The agent can handle basic tasks, but it is not yet ready to completely replace human interaction. OpenAI has had similar experiences with its agent. Users should expect continued refinement before these tools become truly indispensable for everyday web use.