The Promptification of the Search Bar: Evidence of a Behavioral Spillover from Generative AI
This study provides evidence that the widespread adoption of generative AI has caused a durable, population-scale shift in web search query syntax toward more conversational forms—a phenomenon termed "promptification"—which emerged prior to and independently of search platforms integrating generative features, indicating a behavioral spillover where users are reverting to natural language due to their familiarity with LLMs.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
For decades, the way people asked questions of the internet was defined by a strange, silent agreement. When a human needed information, they learned to speak in a compressed, robotic shorthand. Instead of asking a full sentence like "What are the best hiking trails near me?" they typed "hiking trails near me." This habit was not natural; it was a survival skill developed to work with early search engines that could only understand fragments of words. Over time, this keyword style became so ingrained that it felt like the only way to search, even though it required people to strip away their natural curiosity and intent. Then, a new kind of machine intelligence arrived, one that rewarded people for speaking in full, complex sentences. This shift in how we talk to computers has now spilled over, changing the way we use the old search boxes in ways no one expected.
A team of researchers at the Africa Business School in Morocco set out to measure this change. They wanted to know if the conversational habits people learned from talking to artificial intelligence had permanently altered how they typed into the standard Google search bar. To find the answer, they did not rely on surveys or small experiments. Instead, they analyzed a massive dataset of over 196,000 weekly search observations from the United States, covering the period from 2021 to 2025. They looked for specific types of phrases, tracking how often people used simple keywords versus longer, more conversational requests that sounded like they were talking to a person. Their goal was to see if the rise of generative AI, the technology behind tools like ChatGPT, caused a structural break in human behavior, forcing the legacy search bar to adapt to a new way of speaking.
The researchers found a clear and dramatic shift. Following a specific turning point in June 2023, when the use of generative AI began to spread rapidly across the population, the way people typed into search bars changed fundamentally. The frequency of conversational requests—phrases that include words like "help me," "explain," or "can you"—jumped by nearly 59 percent. This was not a gradual drift; it was a sharp, permanent change in the structure of search queries. At the same time, the old style of short, keyword-only searches saw a slight decline. The data showed that this new way of searching did not happen because the search engine company changed its software first. In fact, the shift in human behavior began roughly eleven months before the search engine company itself introduced its own generative AI features into the search bar.
This timeline is crucial because it rules out the idea that the search engine forced people to change their language. If the platform had driven the change, the shift would have happened only after the new features were released. Instead, the data shows that users were already adopting this conversational style while the search bar remained unchanged. The researchers observed that people were bringing the habits they learned from external AI tools back into the traditional search box. They were no longer treating the search engine as a simple index of documents but as a responsive agent capable of understanding complex, natural language. This behavioral spillover suggests that once people experience the ease of talking to a machine in full sentences, they do not easily return to the compressed, keyword-only style of the past.
The study also revealed that this change was not uniform across all topics. The shift toward conversational language was strongest in areas where the information needs were complex and required deep thinking, such as shopping or sports, where people often need synthesis and comparison rather than just a single fact. In these domains, the new, longer search phrases became the standard much faster. The researchers confirmed that this was a genuine, population-wide phenomenon by checking data from around the world, which showed the same pattern. They also used statistical methods to prove that the rise in AI usage was leading the change in search behavior, not the other way around. The data indicates that as more people used generative AI, the search bar naturally evolved to accommodate their new, more natural way of asking questions.
Ultimately, the research documents a moment where human behavior outpaced the technology designed to serve it. The search bar, which had been shaped by the constraints of keyword indexing for decades, is now adapting to a behavior it did not create. People have been freed from the old requirement to speak in fragments, returning to the natural, conversational register that was always their default. The search engine is simply catching up to a new reality where the line between talking to a computer and talking to a person has become increasingly blurred. This study confirms that the influence of artificial intelligence extends far beyond the chat windows where it was first introduced, reshaping the fundamental mechanics of how we seek information in the digital world.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.