Just a few months ago, AI agents struggled to perform even the simplest tasks, such as opening a Chess app on a Mac. However, the all-new ChatGPT super-app has revolutionized the way we interact with our computers.
Released last Thursday, the ChatGPT super-app boasts a range of features, including the built-in Codex coding harness and ChatGPT Work, an AI agent that can access local files and create content in seconds. The revamped ChatGPT also supports "computer use," allowing it to control a PC's mouse and desktop apps.
During a test, ChatGPT successfully opened the Chess app on a Mac mini and began playing a game in under 30 seconds. The AI agent moved the pieces with ease, recognizing the Mac's opening gambit and countering it with a well-executed strategy. This is a significant improvement from previous AI agents, which struggled to navigate the graphical 3D interface.
A similar test with the Claude desktop app showed that it too can use the Mac's Chess app, detecting a different opening move and adjusting its strategy accordingly. However, the ChatGPT super-app's ability to execute tasks with ease and precision makes it stand out from its competitors.
The ChatGPT super-app's computer use capability is not limited to playing Chess. It can also launch and interact with other macOS apps, such as Journal. In one test, ChatGPT opened the Journal app, created a new entry, and began writing a passage about its feelings. The AI agent's ability to navigate and interact with different apps is a testament to its advanced capabilities.
While the ChatGPT super-app's computer use capability is still in its early stages, it has the potential to revolutionize the way we interact with our computers. Developers may find it useful for testing app interfaces and automating repetitive tasks. However, users should exercise caution when allowing the AI agent to access sensitive files and apps.
As the ChatGPT super-app continues to evolve, it will be interesting to see how it is used in the future. One thing is certain: the days of clunky AI "computer use" functionality are behind us, and the possibilities are endless.
