Claude AI is learning how to use a computer like a human.
Anthropic, an AI safety and research company, has released an upgrade to its AI model, Claude 3.5 Sonnet, and a new model, Claude 3.5 Haiku.
The new update includes a significant feature that allows Claude to use a computer like a person. This means that it can navigate a screen, click buttons, and type text.
The new feature, "computer use," is available for developers or public beta. Thanks to this feature, users can ask Claude to do any computer-related tasks for them, such as filling out request forms, analyzing data, looking for information, and more.
Companies such as Asana, Canva, and DoorDash have already used the feature to complete long and repetitive tasks.
And computer use can tackle more than these tasks. It was also built to help develop and test software and conduct research.
This becomes possible due to Anthropic's approach of teaching Claude general computer skills instead of providing it with one tool at a time to complete one individual task.
To make this work, the company built an API that allows the AI tool to interact with the computer interface. Once this API is integrated, it can turn various instructions into commands done on the computer.
The company says that the feature is still being tested, so it may still be tricky to use and may make mistakes. The company also developed classifiers that can tell when computer use is being misused.
Claude 3.5 Sonnet is already available, and Claude 3.5 Haiku will be released sometime later this month.
Your email address will not be published. Required fields are markedmarked