![Anthropic]()
Image: Credit
Anthropic told investors last spring that the company wanted to develop artificial intelligence (AI) to power virtual assistants. These support staff would be able to complete work such as research, email responses, and other office duties on their own. They labeled it a "next-generation algorithm for AI self-teaching" and predicted that it will eventually automate many positions in the economy.
Now, artificial intelligence is becoming a reality.
On Tuesday, Anthropic upgraded its Claude 3.5 Sonnet model. This new model will comprehend and interact with any desktop program. It includes a feature called the "Computer Use" API, which is now accessible for testing. This enables the AI to emulate motions like typing, selecting buttons, and moving the mouse so it can act like a human.
Anthropic noted in a news post, "We trained Claude to see what's happening on a screen and use the features of the software to get tasks done." When developers ask Claude to utilize a software product and grant it access, Claude examines images to see what the user sees. Then, it determines how to shift the pointer so that it clicks in the correct location.
Developers can test the Computer Use feature using Anthropic's API, Amazon Bedrock, or Google Cloud's Vertex AI platform. The updated 3.5 Sonnet model is also available for Claude apps, which improves performance over the previous version.
Anthropic's vision for AI is to create AI-powered virtual assistants for tasks such as research and email responses, with a significant impact on job automation. The enhanced Claude 3.5 Sonnet model now includes a "Computer Use" API for interaction with desktop programs.
Claude can imitate human actions like typing and clicking by analyzing on-screen material. Developers can use this capability through Anthropic's API, Amazon Bedrock, or Google Cloud's Vertex AI platform.