OpenAI's highly anticipated search product, ChatGPT Search, has finally been released, but it's not quite the "Google killer" many expected. After using it as my default search engine for a day, I was left underwhelmed, particularly with its performance on short, navigational queries. These types of searches, which account for the bulk of searches on Google, were often met with unreliable or hallucinated answers.

While ChatGPT Search shines in answering longer, more research-oriented questions, its struggles with short queries make it impractical for everyday use. For example, a search for "Nuggets score" yielded incorrect information, and a search for "earnings today" provided outdated results. These mistakes are a significant hurdle for OpenAI to overcome if it hopes to replace Google as the front door to the internet.

OpenAI acknowledges the struggles with short queries and plans to improve the experience over time. However, until then, Google remains the go-to search engine for many users. The success of ChatGPT Search will depend on its ability to master these short queries, which are essential to its adoption as a daily driver. Despite its shortcomings, ChatGPT Search still offers a glimpse into the potential of AI-powered search interfaces, and its ability to answer longer questions is a promising start.