Large language models like ChatGPT generate answers by querying extensive training data, extending their queries beyond the literal prompt through a method known as “fan-out queries,” a term coined by Google. Understanding these queries allows publishers to grasp the model’s interpretations and priorities better. Two new tools enhance this process specifically for ChatGPT’s web searches in Google Chrome.
The SEO Pub, led by search engine optimizer Mike Friedman, features a free Chrome bookmarklet called the “ChatGPT Search Reasoning and Query Extractor,” which reveals fan-out searches and reasoning. For example, queries for “best headphones for running” include specific product reviews and comparisons. The bookmarklet enables users to access and export this information easily. Another tool, the “ChatGPT Search Fan-out Helper,” saves the fan-out queries along with their corresponding sources, enhancing users’ understanding of the model’s information-gathering approach.
Source link
