The article further discusses the results of an experiment by Dan Taylor, who used ChatGPT as a search engine for 48 hours. He found that ChatGPT has a limited local search experience and often cites the same sources. It also tends to "hallucinate" on some queries. However, it was accurate and timely in citing news sources for live events in sports. The article concludes by stating that while SearchGPT isn't going to disrupt Google, many users are shifting from traditional search engines to LLMs-based chatbots and apps.
Key takeaways:
- ChatGPT is driving approximately 200 million weekly average users to websites and is currently outperforming other search engines in terms of quality and quantity of results.
- Despite its success, ChatGPT has some limitations, such as unreliable results for local searches and a tendency to cite web pages that turn 404.
- ChatGPT, Google and Bing all provide different results for the same queries, with ChatGPT often citing sources not in the top 100 search results of Bing or Google.
- While ChatGPT is not expected to disrupt Google in the near future, many users are shifting from traditional search engines to new AI-based chatbots and apps for their search needs.