Affiliate links on Android Authority may earn us a commission. Learn more.
The first reviews of ChatGPT’s Google Search rival are here, and they’re a mixed bag
- Early reviews of OpenAI’s SearchGPT are trickling in, showcasing a mix of excitement and reservation about its capabilities.
- While some users find SearchGPT effective for specific queries, such as event planning or coding issues, it has its flaws.
- Like other AI models, SearchGPT still suffers from “hallucinations,” sometimes presenting inaccurate information.
OpenAI, the brains behind the wildly popular ChatGPT, has set its sights on the search engine arena with its new contender, SearchGPT. Announced in July, SearchGPT aims to rival Google Search by offering a new AI-driven approach to finding information online. However, a new report by The Washington Post suggests that the tool may not yet be ready to challenge Google’s dominance.
SearchGPT aims to offer a streamlined search experience by delivering concise, organized responses rather than a list of search results. It compiles its answers using data from Bing and other sources, and OpenAI envisions it as a comprehensive resource for users seeking information. Notably, Google also started showcasing AI-generated results above its typical search results, which are called AI Overviews.
The report details user feedback taken from some of the early users of the SearchGPT prototype. Among other shortcomings, users have reported that the AI search bot sometimes provides information that is inaccurate or “hallucinated,” presenting confident but incorrect answers.
One of the reactions shared in the report is from software engineer Ananay Arora, who found SearchGPT’s image results and source labeling disappointing compared to ChatGPT. Others, like tech professional Daniel Lemire, found it superior to Google’s now-infamous AI Overviews but noted that both SearchGPT and Google were less impressive than the AI search startup Perplexity.
AI enthusiast Matt Berman shared his experiences in a YouTube video, highlighting instances where AI search tools outperformed Google on queries about event planning or coding. However, they also encountered cases where SearchGPT provided incorrect details about speakers at an upcoming conference, showcasing the issue of “hallucinated” information. OpenAI has played it smart by launching SearGPT as a prototype first, which should give it enough time to iron out these kinks before a wider rollout.
ChatGPT boasts over 200 million weekly users, many of whom turn to it for questions that might have previously been directed to Google, but it hasn’t significantly impacted Google’s business yet. BrightEdge, a search marketing firm, found that despite its innovative features, SearchGPT still trails behind Google, especially for tasks like online shopping and local searches.
However, given the progress being made, it wouldn’t be unfair to assume that it’s only a matter of time before we encounter a paradigm shift in how we access and interact with information online.