Will answer engines replace search? Four builders on the web's future
Marc Andreessen asks whether AI will replace search, Aravind Srinivas explains why Perplexity builds an answer engine and avoids Google's game, Sam Altman weighs search and ads, and Jimmy Wales considers what language models mean for Wikipedia. Their views show how the way people find information online may change and who stands to lose.
Probably not replace, but displace. Three of the four builders expect direct answers to take over much of what a page of links does today, yet none predicts that search simply disappears. They disagree about what comes next and who pays for it. All four conversations come from a fast-moving period, so the products and business models described may have changed.
Marc Andreessen, who co-created an early browser, gives the broadest forecast. He thinks search as we know it will probably give way to answers, and that AI will treat the internet as its content the way the internet absorbed older media. Links survive because an AI can produce them on request. He also names the supply problem: if nobody visits web pages, people stop making them and a major source of training data dries up.
Aravind Srinivas, the chief executive of Perplexity, is building the thing. His answer engine runs a search, reads the results and writes an answer with every part backed by sources. He does not try to beat Google at ranking links. His bet is on the interface, and on economics: any answer that reduces clicks on links eats into Google's highest-margin business, so Google has little reason to push it. He concedes that Google is still better for reaching a known site and for live information, and says Google is not in danger.
Sam Altman, the chief executive of OpenAI, rejects the contest. Of building a better Google he says "I find that boring". He wants a broader way to find, act on and synthesize information, and says nobody has yet cracked the combination of language models and search. Younger users, he notes, already start with a chatbot.
Jimmy Wales, the co-founder of Wikipedia, expects his site to look much the same in ten years with AI layered on top. He is happy for the knowledge to reach people through assistants. His worry is donations, which depend on users knowing where the knowledge came from.
The sharpest disagreements are over ads and truth. Altman thinks ad support gets worse with AI and prefers that users pay. Wales says "business model drives outcomes" and credits Wikipedia's refusal of ads. Srinivas is open to ads that do not cost user trust. On made-up answers, Andreessen calls them creativity under another name, Srinivas designs against them, Wales says a model does not know or care about the truth, and Altman expects improvement, not a cure. Citations help, but a footnote shows where a claim came from, not that it is true.
What to do with it: use an answer engine for questions that need synthesis, and ordinary search for known sites and live facts. Open the citations and treat unsourced claims as opinion. Keep asking after the first answer. Find out how a tool is paid for before trusting it on a purchase or a contested topic. Check facts separately when they matter. On disputed questions, ask the model to argue each side hard. If you publish online, press for clear attribution. If you use Wikipedia through an assistant, consider chipping in. If you build products, look for the improvement an incumbent's margins discourage.