Ever since OpenAI launched ChatGPT in November 2022, people were offered a new way to get information. You can easily ask the bot questions on various topics, and it will mostly give you reasonable answers. On July 25, OpenAI announced a temporary prototype that includes search features, called SearchGPT. This new feature is intended to be integrated with ChatGPT after the testing phase.
OpenAI CEO Sam Altman said on X that the company believes “there is room to make search much better than it is today.” SearchGPT is designed with real-time access to the internet and brings users the latest information from different sources. For more details, they can follow the links or ask follow-up questions.
SearchGPT is now available to a small group of users and publishers to provide feedback, according to the official press release by OpenAI. Meanwhile, people who want to try the search engine can sign up for the waitlist. Details about the service’s use, like whether it will be paid or when it would be broadly available aren’t known yet.
In shared demo videos, you can see how the prototype works.
OpenAI mentions that SearchGPT provides answers with clear and relevant sources but doesn’t break it down further. One of the factors that may limit the search engine is copyright issues and the concern that media companies may not approve the use of their materials in OpenAI’s search tool. Previously, newspapers including the New York Times, Denver Post, and others already sued the company for using their articles without permission. Though OpenAI built content partnerships with The Atlantic, Time, and others, whether it will use certain resources for its products is still a question.
Google has made moves to integrate AI-based features into its engine to simplify search. In March 2023, the company launched its generative AI chatbot Bard, now Gemini, and in May 2024, announced AI Overviews, a feature powered by their AI model. Google mentioned that the functionality would be available in the US initially, providing overviews for user questions and enabling more organized and simplified search. According to the company, users can get answers to complex questions at once without the need to break down their question into multiple searches.
So far, however, generative AI tools haven’t always been reliable. They can produce incorrect and ridiculous statements, known as AI hallucinations. For example, when Google launched its AI Overviews, users started to share stories where the system suggested eating at least one small rock a day to get minerals and vitamins or adding glue to a pizza recipe to make it “more tackiness.”
SearchGPT is now available to a small group of users and publishers to provide feedback, according to the official press release by OpenAI. Meanwhile, people who want to try the search engine can sign up for the waitlist. Details about the service’s use, like whether it will be paid or when it would be broadly available aren’t known yet.
In shared demo videos, you can see how the prototype works.
SearchGPT results to a user question. Source: openai.com
Here, for example, SearchGPT displays the results for a question about music festivals in North Carolina in August, listing relevant events with short summaries and links to the sources.
SearchGPT vs Google Search: Is Google Losing the Game?
The release of SearchGPT may make competition for Google Search stronger. Among the pros of the tool is that it doesn’t feature ads and is a more interactive way to receive information. Instead of starting a new search, users can ask questions. With GPT 4o’s real-time internet browsing capability, SearchGPT can get any information available on the web. While previously, this privilege belonged to Google and other internet browsers, AI is catching up fast. The question is whether it will be as efficient as Google, and what sources it will use.
OpenAI mentions that SearchGPT provides answers with clear and relevant sources but doesn’t break it down further. One of the factors that may limit the search engine is copyright issues and the concern that media companies may not approve the use of their materials in OpenAI’s search tool. Previously, newspapers including the New York Times, Denver Post, and others already sued the company for using their articles without permission. Though OpenAI built content partnerships with The Atlantic, Time, and others, whether it will use certain resources for its products is still a question.
Talking about Google, its role in the search industry continues to be dominant. According to Semrush, there are 5.9 million Google searches per minute. The browser indexes billions of web pages, which provides users with a wide choice to select their sources of information. Google provides resources and guides on how its algorithms work and how websites are ranked based on their relevance and domain authority. In different cases, however, the search engine may not find the information you need and you would have to open different links and check if the content matches your request.
Google has made moves to integrate AI-based features into its engine to simplify search. In March 2023, the company launched its generative AI chatbot Bard, now Gemini, and in May 2024, announced AI Overviews, a feature powered by their AI model. Google mentioned that the functionality would be available in the US initially, providing overviews for user questions and enabling more organized and simplified search. According to the company, users can get answers to complex questions at once without the need to break down their question into multiple searches.
So far, however, generative AI tools haven’t always been reliable. They can produce incorrect and ridiculous statements, known as AI hallucinations. For example, when Google launched its AI Overviews, users started to share stories where the system suggested eating at least one small rock a day to get minerals and vitamins or adding glue to a pizza recipe to make it “more tackiness.”
How accurately SearchGPT will perform in this aspect is not yet clear, but an article from The Atlantic mentions that the engine made a mistake in a demo video, incorrectly recommending a user to attend a festival that had ended in July as if it's about to happen in August.
On the other hand, traditional search can also feature incorrect or confusing information in the top results. You may need to cross-check the results with original sources to get the most accurate information possible. Besides, search results often include lots of ads, especially when you're looking for expert advice or a product or service to use. Some of the ads can direct you to low-quality websites or, worse, scam pages that request your card number without providing a valid service. Analysis by The Wall Street Journal states that AI tools made it easy to produce content and many are exploiting this opportunity to promote their pages on the web. Often, they use platforms favored by Google, such as LinkedIn, Reddit, and Quora to improve their rankings. So when comparing AI and traditional search, there is no perfect search option to use yet, as both may have errors. This means that whichever method you choose, you need to be careful with the sources and information provided, verifying the website domains and ensuring that the authors presenting themselves as experts are not faking their identities or expertise.
Competition among online search tools is expected to grow, offering users a broader range of options. According to Gartner, by 2026, traditional search engine volume will drop by 25% due to generative AI chatbots and virtual assistants. Meanwhile, search engines will increase their focus on featuring quality content in organic search results.
Different search methods will allow users to find the tools that best meet their needs. While AI bots may excel at providing quick answers, traditional search engines can offer a wider array of sources and diverse content styles.
On the other hand, traditional search can also feature incorrect or confusing information in the top results. You may need to cross-check the results with original sources to get the most accurate information possible. Besides, search results often include lots of ads, especially when you're looking for expert advice or a product or service to use. Some of the ads can direct you to low-quality websites or, worse, scam pages that request your card number without providing a valid service. Analysis by The Wall Street Journal states that AI tools made it easy to produce content and many are exploiting this opportunity to promote their pages on the web. Often, they use platforms favored by Google, such as LinkedIn, Reddit, and Quora to improve their rankings. So when comparing AI and traditional search, there is no perfect search option to use yet, as both may have errors. This means that whichever method you choose, you need to be careful with the sources and information provided, verifying the website domains and ensuring that the authors presenting themselves as experts are not faking their identities or expertise.
Is Generative AI the New Way to Search?
AI's role in search is still experimental. Both established and new search tools, such as SearchGPT, Google Search, and Perplexity, are working to optimize the search process with AI. Perplexity, launched in 2022, was among the first to represent an interactive search through generative AI. It soon gained popularity due to its user-friendly interface and combination of traditional search with generative AI. The website's home page features a box saying "Ask anything" where you can type your questions and get answers for free. As of the latest data, Perplexity has 10 million monthly active users and processed half a billion queries in 2023. The startup has raised over $120 M from NVIDIA, Jeff Bezos, Nat Friedman, and others.
Competition among online search tools is expected to grow, offering users a broader range of options. According to Gartner, by 2026, traditional search engine volume will drop by 25% due to generative AI chatbots and virtual assistants. Meanwhile, search engines will increase their focus on featuring quality content in organic search results.
Different search methods will allow users to find the tools that best meet their needs. While AI bots may excel at providing quick answers, traditional search engines can offer a wider array of sources and diverse content styles.
Anahit Avetisyan
Author
Web3 writer and crypto HODLer with a keen interest in market trends and recent technologies.
