OpenAI announced the launch of an agent artificial intelligence (AI) called Operatorwhich can go to the web to perform tasks for the user. This way, you could look at a web page and interact with it by typing, clicking, and scrolling. Initially, it is available for ChatGPT Pro level subscribersin the United States.
The operator of the company led by Sam Altman is based on a model of “computer-using agent” which combines the vision capabilities of GPT-4o with “advanced reasoning through reinforcement learning” to be able to interact with the graphical user interface (GUI), OpenAI detailed.
“The operator can ‘see’ (via screenshots) and ‘interact’ (using all the actions that a mouse and keyboard allow) with a browser, allowing him perform actions on the web without the need for custom API integrations“, according to OpenAI.
Likewise, the operator it can use reasoning to “self-correct” and if it crashes it will give control to the user. It will also prompt the user to take control when a website requests sensitive information, such as login credentials, and “should” request that a user approve actions such as sending an email.
In that sense, OpenAI also explains that Operator was designed in order to “reject harmful requests and block disallowed content.”
The company also states that it is collaborating with companies such as DoorDash, Instacart, OpenTable, Priceline, StubHub, Thumbtack and Uber so Operator “addresses real-world needs while respecting established standards”.
However, the company warns that everything may not work as expected for the moment; the tool currently has issues with “complex interfaces like creating slideshows or managing calendars.” Finally, they assured that they plan to bring this new tool to Plus, Team and Enterprise users and “integrate these capabilities into ChatGPT.”
ChatGPT returned to work after its global crash
ChatGPT, The most used artificial intelligence (AI) chatbot, returned to work and recovered its services after the global fall it suffered, as reported by millions of users who use it daily for work, study, or simply for everyday actions.
When trying to log in, users received the message “internal server error”. This is not the first time they have experienced problems: on December 26, 2024, the system experienced failures.
OpenAI stated on their official page that they were “experiencing high error rates in the API. A fix has been implemented and we are monitoring the results.” On previous occasions, during similar outages, the company contacted users to provide details about the problems they were facing.
Source: Ambito
I am Pierce Boyd, a driven and ambitious professional working in the news industry. I have been writing for 24 Hours Worlds for over five years, specializing in sports section coverage. During my tenure at the publication, I have built an impressive portfolio of articles that has earned me a reputation as an experienced journalist and content creator.
