AI researchers advise selective use of chatbots for web searches

In a wave of guidance that could reshape how millions look up facts online, leading AI researchers and executives from Google, Microsoft and OpenAI said on Tuesday that chat‑based assistants should be used selectively, not as a blanket replacement for traditional search engines. The statement followed a joint report that compared the performance of large‑language‑model chatbots with conventional search tools across a set of everyday queries. The researchers emphasized that the choice of tool can affect both speed and accuracy of the information retrieved.

The advice comes as AI chatbots have moved from niche experiments to mainstream products. Since the launch of ChatGPT in late 2022, services like Bing Chat and Google’s Gemini have been embedded in browsers and mobile apps, promising answers in conversational form. Users are attracted by the ease of typing a question and getting a single, synthesized response. At the same time, search engines have refined algorithms that rank pages based on relevance and authority, a process that still underpins most online research. The clash between these two approaches has sparked debate about reliability, transparency and the risk of misinformation.

A series of independent tests cited in the report found that chatbots answered correctly about 68 % of factual questions, while top‑ranked search results were accurate in roughly 85 % of cases. The gap widened for niche topics, where chatbots often hallucinated details that did not exist on any web page. Users who tried both methods reported that search engines were faster for simple look‑ups, whereas chatbots saved time when they needed a quick summary or a list of options. Industry analysts warned that over‑reliance on AI could erode critical evaluation skills, especially among younger users who may trust a polished response more than a list of sources.

Going forward, the companies involved said they will add clearer labels to AI‑generated answers and improve citation mechanisms that point users to the original web pages. Regulators in the EU and US are watching the rollout of these features, considering rules that would require transparency about when a response is AI‑crafted. For everyday users, the recommendation is simple: ask a chatbot when you need a concise overview, but switch to a search engine when you need the most up‑to‑date facts or want to verify the source. The balance between convenience and accuracy will likely shape the next stage of online information seeking.