OpenAI's SearchGPT has mixed feedback from early users. While some users have suggested that the platform's search results provide incorrect information, others have praised its ability to deliver precise answers over Google AI Overviews. 

Launched in July, SearchGPT aims to provide organized, concise responses rather than traditional search result lists. OpenAI has previously claimed that SearchGPT was designed to speed up searches and provide direct, relevant answers with clear source citations. However, users’ feedback highlights several areas where SearchGPT falls short.

Inaccuracies and “hallucinations”

Early feedback from users reports that SearchGPT sometimes provides incorrect or “hallucinated” information. 

This issue was highlighted by Software Engineer, Ananay Arora, who found the tool's image results and source labeling lacking compared to ChatGPT and Google according to The Washington Post. Arora said that while SearchGPT provides accurate results for local restaurant searches, it’s generally unimpressed with its images and sometimes confusing source labels for other queries. 

BrightEdge’s Founder, Jim Yu, however, estimated that less than 1% of searches in SearchGPT returned incorrect information as per The Washington Post.

Competitor comparison

While some users, such as Tech Professional, Daniel Lemire, found SearchGPT superior to Google’s AI Overviews, it still falls short of the new AI search startup Perplexity AI.

He claimed that SearchGPT responses provide better results than Google AI Overviews results. However, responses from Perplexity AI are more effective than those provided by SearchGPT and Google AI Overviews.

AI Enthusiast, Matt Berman's experiences highlighted that SearchGPT outperformed Google in some specific queries like event planning and coding. He also highlighted that SearchGPT struggled with accuracy issues, such as providing incorrect information about event speakers.

Market comparison and position

The Washington Post further explained that despite its features, SearchGPT is currently trailing behind Perplexity AI and Google AI Overviews, especially in areas like online shopping and local searches.

A study by SE Ranking found that when compared with Google and Bing, SearchGPT's search results align 73% with Bing's but only 46% with Google's. 

The study also found that SearchGPT provides fewer and less diverse results than Google but “it lacks additional elements like maps, videos, or different useful SERP features that Google has.”

According to the study, while “SearchGPT is best for informational queries because it delivers detailed, high-quality summaries from various sources,” it favors authoritative sources more. It also fails to provide up-to-date results compared to Google. 

SearchGPT is also effective at generating helpful lists and comparisons for commercial queries but it does not handle searches like navigational, local, and transactional queries best.

SearchGPT is currently in its prototype stage. OpenAI Spokesperson, Kayla Wood, told The Washington Post that the platform will use the feedback collected from early users to make adjustments to the prototype before its full release.