OpenAI has launched the latest AI proxy "Operator", a powerful tool designed to help users perform various tasks online. It combines GPT-4o's visual capabilities and advanced reasoning for reinforcement learning, and is able to interact with graphical user interfaces (GUIs) and act independently on the network without custom API integration. Operator is currently in the research preview stage and is only available to US ChatGPT Pro subscription users, with a monthly fee of US$200. This article will introduce the functionality, security and future development plans of Operator in detail.
OpenAI announced the launch of its latest AI agent "Operator", a tool designed to help users perform various tasks on the network. OpenAI said in its blog that Operator is conducting a "research preview" and is initially targeting ChatGPT Pro subscription users in the United States, with a monthly fee of $200.
Operator’s design philosophy is to interact with a graphical user interface (GUI) through a model called “Computer Usage Agent” that combines GPT-4o’s visual capabilities and advanced reasoning with reinforcement learning. OpenAI explains that Operator can view web pages through its built-in browser and interact with the pages by typing, clicking, and scrolling. The advantage of this technology is that Operators can operate independently on the network without the need for customized API integration.
During use, Operator not only uses reasoning ability to "correct itself", but also returns control to the user when encountering difficulties. When a website requests sensitive information, such as login credentials, the Operator asks the user whether to take over the operation. In addition, Operator also requires users to confirm when handling transactions such as sending emails. OpenAI emphasizes that Operators are designed with special emphasis on security, aiming to reject harmful requests and block unauthorized content.
OpenAI also revealed that Operator is working with several well-known companies such as DoorDash, Instacart, OpenTable, Priceline, StubHub, Thumbtack and Uber to ensure they meet real-world needs and follow established industry norms. However, OpenAI also reminds users that the tool may currently experience difficulties when dealing with complex interfaces, such as creating slideshows or managing calendars.
OpenAI plans to expand Operator to Plus, Team, and Enterprise users and integrate these capabilities into ChatGPT. This means that more users will have the opportunity to experience the convenience brought by this cutting-edge technology.
Official podcast: https://openai.com/index/introducing-operator/
Points:
OpenAI launches the "Operator" AI agent to help users perform tasks online, and is the first to target ChatGPT Pro users.
Operator can interact with web pages through the browser, and has the functions of self-correction and user control to ensure security.
OpenAI cooperates with many well-known companies to meet real needs, while also planning to expand to more users in the future.
In short, Operator is a bold attempt by OpenAI in the field of AI proxy, which demonstrates the great potential of AI in automating network tasks. Although it is still in its early stages, its future development is worth looking forward to and also indicates more possibilities for AI to interact with humans in the future.