OpenAI CEO Sam Altman is interested in a different kind of AI search: one built around language models, information synthesis and user intent, rather than a direct attempt to recreate Google. In an interview with podcaster Lex Fridman, Altman framed the opportunity as a broader rethink of how people use the Internet to get answers and take action.
Altman Wants To Rethink Search, Not Copy Google
Altman described today’s dominant search experience as something OpenAI does not simply want to imitate. The point, in his telling, is not to produce a cleaner or more competitive version of the familiar search results page.
Instead, he said the more interesting challenge is building a system that helps users work with information more directly. That could mean finding material, turning it into something useful, comparing it, summarizing it or helping a person decide what to do next.
"The thing that’s exciting to me is not that we can go build a better copy of Google search, but that maybe there’s just some much better way to help people find and act on and synthesize information," Altman said.
That framing matters because it places AI search in a different category from traditional web search. Google search is built around retrieving and ranking links. Altman is pointing toward an experience where the system may generate an answer in some cases, summarize information in others, or provide references when that is the better path.
ChatGPT is already part of that shift. According to the source, Altman sees ChatGPT as an example of some alternative use cases, and OpenAI hopes to expand those capabilities into many more situations.
The Search Problem Is Technical, Strategic And Cultural
Altman also acknowledged that building a better search service than Google is a huge challenge. The difficulty is not only technical. It also involves branding and the broader ecosystem around search.
That is one reason his comments are not a simple declaration that OpenAI wants to replace Google with a similar product. He explicitly rejected that approach.
"I don't think the world needs another copy of Google," Altman said.
Asked what would be more compelling than a standard Google clone, Altman pointed to the intersection of language models and Internet search. He said OpenAI is interested in how to combine those pieces well.
"That would be an example of a cool thing," Altman said. "The intersection of LLMs plus search, I don't think anyone has cracked the code on yet. I would love to go do that. I think that would be cool."
The phrase "cracked the code" is important here. It suggests that AI search is still unresolved, even as companies explore it aggressively. A useful AI search system would need to be fast, reliable and clear about when it is answering, summarizing or referencing information.
Why Ads Are Central To The AI Search Debate
Altman also raised a business model question: whether AI search needs to rely on advertising. He said OpenAI does not need to do so for aesthetic reasons, and he described advertising on the Internet as something that was necessary for various reasons to get the Internet off the ground, but also as a temporary phenomenon.
His preference is tied to the way ChatGPT is paid for by users. Altman likes that users can pay for ChatGPT and understand that its answers are not shaped by advertisers.
That concern becomes sharper when search becomes conversational. If an AI assistant recommends a product, a trip or a service, the user needs to know whether that answer is based on their needs or on commercial pressure. Altman said it is easy to imagine a dystopian future in which ChatGPT tells a user to buy this or that product or take a vacation somewhere.
In traditional search, ads are visible as part of the page experience. In AI search, the line between recommendation, answer and promotion could become more sensitive. Altman’s comments suggest that OpenAI sees user trust as part of the product, not just a design detail.
OpenAI May Be Building Toward A Dedicated Web Search Product
The source reports that OpenAI is developing its own web search product to compete with Google. The product is said to be partly based on Microsoft's Bing search. It is not yet clear whether the web search product would stand apart from ChatGPT or be integrated into Bing.
That uncertainty reflects a broader question about how AI search should be packaged. It could live inside ChatGPT as a capability. It could become a separate search service. It could also connect to Microsoft’s existing search ambitions.
The source notes that ChatGPT's current web search is still slow and unreliable, which could make a separate, optimized search product from OpenAI sensible. Microsoft is also interested in taking market share away from Google in the search engine market.
Microsoft CEO Satya Nadella’s vision of using chatbots that automatically search the web has not yet been realized. That leaves room for a different implementation, especially if OpenAI can make search feel less like a link list and more like a task-focused assistant.
AI Agents Could Extend Search Into Action
The source also says OpenAI's web search product could be linked to an AI agent that independently performs tasks on the Internet, such as reserving movie tickets. That would move search beyond information retrieval and into execution.
In that model, the user would not only ask for information. They might ask the system to help complete a practical task. That makes accuracy, reliability and accountability more important because the system could affect real-world outcomes.
Google also sees a possible agent future as a complement to traditional search, according to the source. At the same time, Google currently has what the source describes as probably the best AI search approach with its Search Generative Experience.
Even so, many unresolved questions remain. The source points to higher costs, liability for AI answers, whether providers of such services might become media companies subject to corresponding rules and laws, and how this would change competition with other media companies.
For Google, there is another tension: the company is likely to have little interest in jeopardizing its existing search engine monopoly. For OpenAI, that creates an opening. Altman’s comments suggest the target is not a familiar search page with an AI layer attached, but a search experience shaped from the start around language models, synthesis and action.