OpenAI, the maker of ChatGPT, launched Operator on Thursday. This is an assistant, or in technical terms an agent, who can use artificial intelligence to perform tasks on the internet for the user, such as planning a holiday, booking a restaurant or shopping. OpenAI thus joins the race for increasingly powerful AI assistants.
Operator uses his own browser, can look at a website, scroll, press buttons and fill in text fields as people do every day, the Californian company said in a press release. The new feature is currently only available to ChatGPT business subscribers to improve them based on feedback.
“Operator is one of our first agents, in other words, AI that is able to autonomously perform tasks for you: you give a task and Operator executes it,” said OpenAI.
Race
The rapid growth of generative AI, with the success of ChatGPT since late 2022, has sparked a frantic race for AI assistants among the tech giants. They are constantly launching tools that can write messages, answer questions, and generate images, among other things.
The Holy Grail of Silicon Valley is AI agents, a kind of omniscient secretary that is always available and can perform a variety of tasks, from sending messages to running errands on the Internet.
OpenAI is not the fastest. Operator is similar to Computer Use, a feature launched in October by Anthropic, a rival startup. Google, which launched Gemini 2.0, the new version of its AI model, in December, is also working on more complex interactions with technology so that AI agents can surf the web on their own and search for additional information online or in a document.
All companies indicate that their AI agents act under the supervision of people: they can put products in the shopping cart in an online store, but cannot (yet) click on the payment button.