The Race to Build a ChatGPT-Powered Search Engine
In this story, the author looks at how companies like Microsoft, Google, and others are scrambling to bolt ChatGPT-style language models onto search engines. The goal is to offer answers in natural language instead of long lists of links.
Why conversational search is attractive
Language models can summarize many sources at once, write in a friendly tone, and handle follow-up questions. That is appealing for users who are tired of digging through SEO-stuffed web pages. The article explains how this could make search feel more like a conversation with a knowledgeable assistant.
The stubborn problems under the hype
At the same time, the author stresses the costs and risks. Running large models for every query can be far more expensive than traditional search. Worse, these systems still hallucinate—confidently inventing facts—and can inherit biases from their training data. That creates a tension between convenience and trustworthiness.
Uncertain business models
Finally, the article points out that the classic search business model is built on showing ads next to results. It is not obvious how advertising should work inside a chat interface, or whether people will accept it. The race to build AI-powered search is real, but the economic and technical foundations are still unsettled.
Read the original article on WIRED.
