OpenAI launches Operator, a new AI agent to browse and interact with websites via ChatGPT
OpenAI has introduced a "research preview" of Operator, an AI agent designed to perform web tasks for users completely autonomously. The AI uses a "Computer-Using Agent" model that combines GPT-4o's vision capabilities with advanced reasoning through reinforcement learning, allowing it to interact with graphical user interfaces, which means Operator can interpret web pages via screenshots and execute actions like typing, clicking, and scrolling through an embedded browser in ChatGPT, without the need for custom API integrations.
This new feature will let users complete complex tasks on specific websites by simply asking ChatGPT, such as booking a family vacation for a specific date on Booking.com or finding a product with specific features on Etsy features self-correcting abilities and can transfer control to users if it encounters difficulties. It includes safety measures to refuse harmful requests and protect sensitive information, seeking user approval for sensitive actions like entering login credentials.
Available initially in the United States through operator.chatgpt.com, Operator is exclusive to ChatGPT Pro subscribers at a cost of $200 per month. OpenAI plans to expand Operator's availability to Plus, Team, and Enterprise tiers and integrate it further into the platform.
- AI Chatbot
- Freemium
- Proprietary
Related news
External links
- Introducing OperatorOpenAI Blog • Official source
- OpenAI's Thread on X@OpenAI on X • Official source
- OpenAI Operator agent can automate tasks such as vacation planningCNBC
- OpenAI launches Operator, an AI agent that performs tasks autonomouslyTechCrunch
- OpenAI’s new Operator AI agent can do things on the web for youThe Verge
Comments
On the same boat with hatena8107 here in that it's a great innovation, most likely something that's been in the works for a year or two; but at the same time, it shouldn't be something that'll take away more of the human work away from you. There's a reason it's only available on the $200/month Pro plan (for now).
And privacy-wise, I'm not so fond of ChatGPT, but like it or not, it is getting smarter everyday. At least from what I'm seeing anyway.
Yes indeed. Let us embrace laziness. The future of humankind is sitting on your couch while AI does things for you. Using data aggregated from our partners, our AI will decide the optimal products, places, relationships and entertainment - For You. For just $200 a month, get access to this incredible technology, and advance your human selves evermore!
rant No, really though. How much more is there to take away from humans? So far they've got relationships (AI sexting, friends, etc), reading information for you (AI summaries of websites, documents, product ratings...), writing things for you (documents, texts, letters...), creating things for you (music, writing, art...), buying things and choosing what to watch for you (this has existed for a while through "recommendations", though...). It wouldn't take much for companies to start buying things for you by analyzing what you buy and at what time, time between product purchases, etc. and that's not factoring in your personal life (social media, what happens in the home, etc). You wouldn't even be able to use any of the products they buy for you, because you'll be working full-time trying to make all the money for them to purchase those goods, never mind all the streaming services they'll subscribe you to because you have an 86% chance of being interested in this and that show. /rant
Don't take my stance on modern AI the wrong way: I think it's a marvelous innovation. It's good when it's used as a tool. The intended use put forth by the owners of these AI is what upsets me. Replacing human autonomy, or marketing in such a way as to reduce human autonomy for "convenience". It's simply monetizing laziness, or perhaps even the ever-growing lack of time for things in this interconnected world...but that's another can of worms entirely.