Generative artificial intelligence heavyweight OpenAI on Thursday previewed an AI agent capable of performing tasks on the web for users, as it seeks to improve its chatbot amid increased competition .
The tool, called Operator, is powered by a template that allows it to interact with buttons, menus and text fields on the screen.
“This capability marks the next step in the development of AI, enabling models to use the same tools that humans use every day and opening the door to a vast range of new applications,” the company said in a post of blog.
Operator can perform a range of tasks, such as creating to-do lists or helping with vacation planning. It also takes user data into account when it judges a task to be completed and asks for confirmation for certain tasks, such as entering login details on a website.
The tool is currently available to Pro users in the United States as a research preview, the Microsoft-backed startup said.
-Agents, which are systems capable of carrying out actions such as making purchases or scheduling meetings without direct human intervention, are at the forefront of businesses’ AI concerns today.
Perplexity, an OpenAI competitor, launched an agent-based assistant for Android devices earlier on Thursday. This assistant can, among other things, reserve dinners, hail passengers on applications and schedule reminders.
Last year, Apple integrated Apple Intelligence into its voice assistant, Siri, and — in a partnership with OpenAI — the iPhone maker also introduced the use of ChatGPT, with the user authorization.
While such agents have long eluded researchers, the emergence of step-by-step reasoning approaches like those used in OpenAI’s o1 model could make such tasks possible, company executives told Reuters in December.
