OpenAI's "GPT Search" trial feedback: Far from replacing Google, not as good as Perplexity
The "GPT Search" launched by OpenAI has received less than ideal feedback during the trial phase, with users expressing disappointment in its interface and answers, especially in handling images and source annotations. Furthermore, although GPT Search is positioned as an alternative to traditional search tools, it has not yet posed a threat to Google's market dominance, and it performs poorly in some queries, potentially providing inaccurate information
OpenAI is developing a search engine called "GPT Search," which is dubbed as the "Google Killer," but it seems to be not quite up to par at the moment.
In July this year, OpenAI announced the launch of GPT Search, further sparking market interest. In recent weeks, OpenAI has started providing prototypes to 10,000 early testers.
This search engine is able to use AI to provide concise answers with clear sources by aggregating information from different web pages. However, it appears to be "underpowered" when dealing with certain shopping and local queries, sometimes even presenting unreal or "illusory" information.
Lagging Behind Google, Inferior to Perplexity
In the current version of GPT Search, OpenAI positions GPT Search as "a one-stop shop for any information people are looking for, rather than a traditional search tool that provides clickable options."
However, testers have mixed reviews of GPT Search. Software engineer and AI and cybersecurity researcher Ananay Arora stated that while he was satisfied with the results for local restaurant queries, he was disappointed with the image search results and occasionally found the labeling of "sources" confusing.
A report released by BrightEdge shows that Google still leads in online shopping or local information searches, providing access to specialized data to answer queries such as "business listings, hotel room availability, and flight information." However, obtaining such data typically comes at a high price. In 2011, Google spent $676 million to acquire a flight data provider to optimize its travel search results.
On the other hand, Daniel Lemire, head of the educational organization AI Mistakes, gave a more positive evaluation of GPT Search, believing that OpenAI's search tool is better than Google's AI-generated answers or "summaries" added to its search results page.
Nevertheless, both Arora and Lemire believe that:
Compared to a small startup named Perplexity, OpenAI's interface and answers are not quite impressive enough.
This emerging competitor Perplexity uses AI search networks to provide one-time answers with source links, allowing users to switch modes manually and choose models in advanced versions.
GPT Search is still in its initial prototype stage
One of the main challenges faced by AI search tools is the issue of providing inaccurate or "illusory" information. In July, in a video demonstration of the launch of GPT Search by OpenAI, the engine provided incorrect music festival dates AI enthusiast Matt Berman also believes:
"One big drawback of AI search is that it confidently tells you some completely wrong things."
Wood from OpenAI stated: "GPT search is an initial prototype, and we will continue to improve it."
Facing the upcoming election, OpenAI stated that ChatGPT will respond to questions about election procedures by guiding users to authoritative information.
Despite its limitations, Jim Yu, founder and CEO of BrightEdge, still believes that GPT search has the potential to disrupt the way people access information online, much like the iPhone disrupted mobile computing.
Has "GPT Search" Failed to Continue the Brilliance of ChatGPT?
OpenAI's ChatGPT once sparked predictions that AI chatbots would completely change the way people search for information online, potentially marginalizing some Silicon Valley giants and shaking Google's decades-long dominance in the search engine market.
However, its search engine has been somewhat disappointing. According to user feedback and analyses from search marketing companies, this "GPT search" is far from posing a threat to Google.
Last month, OpenAI stated that ChatGPT is used by over 200 million people on average per week, with many asking questions that may have been previously entered into Google.
Previously, OpenAI spokesperson Kayla Wood mentioned that GPT search will "incorporate the best features into ChatGPT".
However, some testers have expressed dissatisfaction with certain features of GPT search:
"Considering that companies like OpenAI have a history of having the most advanced models, you would expect them to make breakthroughs, but compared to ChatGPT, GPT search is not very impressive."
It is worth noting that according to search marketing company BrightEdge, Microsoft added a chatbot powered by OpenAI technology to its Bing search engine last year, but the company's market share only slightly increased