OpenAI is constantly working for the next major development to add to ChatGPT, and following months of rumors, including a piece from earlier this week that hinted at a launch, the technology giant’s first AI Agent has arrived. The operator is designed to accomplish web chores for you with a single button press.
The Operator is a Computer Using Agent (CUA) that employs GPT-4o’s visual abilities to navigate and search the internet. This means that it can grasp the context of what to search for, as well as what it sees when searching, thanks to its multimodality. It’s now accessible as a research sample for ChatGPT Pro subscribers in the United States.
The operator is defined as “an agent which can use its web browser to complete tasks for you.” OpenAI posted a demo of Operator perusing the web as we (humans) do. You could ask the Operator to make reservations for dinner for you, fill out a lengthy paperwork, get goods from a service, or even arrange a ticket. It can use OpenTable to find and schedule a restaurant reservation, as seen in the demo. The operator will even walk you through the steps.
Operator is a research preview,’ therefore be aware that it is in its early stages. OpenAI does have certain limitations. We haven’t had a chance to try it out yet, but it looks impressive. This is OpenAI’s first foray into the world of AI agents, which will most likely be the theme of the year in the field of artificial intelligence.
The operator is powered by the new Computer Using Agent (CUA) model, which combines GPT4o’s visual capabilities with advanced thinking. This all comes together to allow the Operator to understand and interact with browser features such as the search bar, various buttons, and on-screen content.
According to OpenAI, “Operator can see (through screenshots) and ‘interact’ (using all the actions a mouse and keyboard allow) with a browser,” allowing it to execute a task functionally. That’s really cool, especially if it works with a high percentage of success and, according to the blog article, can self-correct.
ALSO READ: Canon Unveils World’s First 410MP Full-Frame Sensor for Industry
