OpenAI is overhauling the way ChatGPT responds to prompts, ditching plain blocks of text in favor of interactive software generated on the spot.
The update, titled Intelligent UI, gives the chatbot the ability to generate custom visual components when an answer calls for more than paragraphs. Built on the company's GPT-6 model, the new experience began rolling out to paid subscribers on Wednesday. Accounts on the ChatGPT free tier will receive the feature on Thursday.
Rather than forcing users to read through paragraphs or scan static pictures grabbed from the web, the system codes bespoke mini-apps inside the conversation stream. In pre-launch tests reported by Wired, prompts generated tactile interfaces rather than plain copy. A request for slug anatomy produced a diagram with dedicated buttons for sections like the muscular foot and upper tentacles. A search for rental costs in San Francisco yielded an affordability calculator with operational sliders for expenses and wages, plotting out the exact share of take-home pay required for various apartments.
Another demonstration produced an interactive cabin explorer. Asked to contrast four aircraft flown by Delta and Alaska, the bot generated a side-by-side seating map. Users could tap individual seats to read details, with exit rows marked in green.
Dialing Back the Clutter
OpenAI is trying to keep the conversational screen from turning chaotic. Not every question will spit out a chart or a cluster of buttons.
"We have a design team that spent a lot of time figuring out when adding a diagram, chart, or buttons adds value versus when it's starting to feel cluttered," said Aarush Selvan, a product manager at OpenAI. Users who prefer a standard reading experience can simply prompt ChatGPT to provide fewer visual components.
The redesign puts OpenAI on a parallel track with Google. Earlier this year, Google introduced its own generative UI within Google Search, creating interactive graphics with adjustable sliders to explain complex science topics like black holes.
For ChatGPT's user base of more than one billion people, the update marks a distinct break from the chat window's rigid text origins. As generative tools weave together text, images, and voice into flexible software interfaces, the layout of an application is increasingly assembled for one specific prompt. Free users will get their first look at the new setup starting Thursday.
ADFiled by The AI Desk
Models, assistants and the companies and chips behind them, reported from what was released and what was claimed, with the difference kept clear.
More from this desk →
Be the first to comment
Join the argument. No password, just your email or a passkey.