OpenAI is making waves with its highly anticipated foray into the search engine market with SearchGPT, an innovative AI-driven search engine that offers real-time access to information across the internet.
SearchGPT greets users with a large textbox asking, “What are you looking for?” Unlike traditional search engines that provide a plain list of links, SearchGPT aims to organize and make sense of the information it retrieves. For instance, when searching for music festivals, the search engine not only summarizes the findings but also presents short descriptions of the events, followed by attribution links.
In another demonstration, SearchGPT details the optimal times for planting tomatoes and breaks down different varieties of the plant. After presenting the initial results, users can ask follow-up questions or explore other relevant links via a sidebar.
Currently, SearchGPT is a “prototype”. Powered by the GPT-4 family of models, it will initially be accessible to only 10,000 test users, according to OpenAI spokesperson Kayla Wood. OpenAI is collaborating with third-party partners and utilizing direct content feeds to enhance its search results. The ultimate goal is to integrate these search features directly into ChatGPT.
This move by OpenAI could pose a significant challenge to Google, which has been swiftly integrating AI features into its search engine to prevent users from migrating to competitors with advanced tools. It also positions OpenAI in direct competition with Perplexity, an AI “answer” engine that has faced criticism for allegedly ripping off publishers’ work with its AI summaries feature.
OpenAI appears to have learned from Perplexity’s missteps. In a blog post, the company emphasized that SearchGPT was developed in collaboration with various news partners, including the owners of The Wall Street Journal, The Associated Press, and Vox Media. “News partners gave valuable feedback, and we continue to seek their input,” Wood stated.
Publishers will have the option to manage their appearance in OpenAI’s search features. They can opt out of having their content used to train OpenAI’s models while still appearing in search results. “SearchGPT is designed to help users connect with publishers by prominently citing and linking to them in searches,” according to OpenAI’s blog post. “Responses have clear, in-line, named attribution and links so users know where information is coming from and can quickly engage with more results in a sidebar with source links.”
Releasing SearchGPT as a prototype offers OpenAI several advantages. It provides a buffer if the search results are inaccurate — similar to Google’s AI Overviews suggesting users put glue on pizza. Additionally, it mitigates potential issues like incorrect attributions or the unintentional replication of content, as seen with Perplexity. The development of this new product has been a topic of speculation for months. Reports from The Information in February and Bloomberg in May have chronicled its progress. Concurrently, OpenAI has been actively recruiting Google employees for its search team, and observant X users have noticed a new website hinting at this launch.
OpenAI has been steadily connecting ChatGPT with the real-time web. When GPT-3.5 was released, its data was already months old. Last September, OpenAI introduced a feature called Browse with Bing, which allowed ChatGPT to browse the internet, though it was less sophisticated than SearchGPT.
The rapid advancements by OpenAI have garnered millions of users for ChatGPT, but the associated costs are mounting. The Information reported that OpenAI’s AI training and inference costs could reach $7 billion this year, with the free version of ChatGPT further driving up compute costs. SearchGPT will be free during its initial launch, and with no ads currently, OpenAI will need to devise a monetization strategy soon.
