The ChatGPT search engine is coming, but it still faces a lot of trouble

Interface News Reporter | Chen Zhenfang

Interface News Editor | Song Jianan

OpenAI, the new AI hegemon, has begun to challenge the search market landscape.

OpenAI is likely to launch a new search engine based on ChatGPT technology. On May 6, Jiemian News noticed that a webpage called "GPT Search" had been launched, but it is currently only accessible to members.

The above news was first revealed by a post in the startup incubator Y Combinator, which stated that the domain name and security certificate for the "search.chatgpt.com" had been created. Well-known journalist Pete Huang also tweeted a trailer, saying that GPT Search will be officially launched on May 9.

How OpenAI can provide a more personalized and deeply integrated search experience, and what kind of change in the way people interact with information is highly anticipated. Since the birth of ChatGPT, the tech community has been waiting for a head-to-head showdown between OpenAI and Google in the search market. A large number of users have already obtained AI-generated search answers through chatbots such as Microsoft's Bing Chat and Google's Gemini.

Currently, OpenAI is valued at more than $80 billion, and Microsoft has invested $13 billion in the company. OpenAI appears to be "ambitious" in the search market.

"With deep integration of LLMs (large language models) with search, ChatGPT could be the future of search. In April, OpenAI's CEO Sam Altman made the company's ambitions clear.

The company, which holds a $13 billion investment from Microsoft, is currently valued at more than $80 billion. Technically and financially, it has become a new force that can threaten Google's search position.

In fact, OpenAI is already using the power of Microsoft to keep moving closer to search engines. Last May, Microsoft announced at its "2023 Build Developer Conference" that Bing search would soon be integrated directly into OpenAI's ChatGPT chatbot, allowing some real-time search capabilities to be built into Bing.

Microsoft also said that ChatGPT will have a world-class search engine built into it to provide more timely and updated answers and get information from web pages.

At the time, Satya Nadella, CEO and chairman of Microsoft, was optimistic that "the entire search category is undergoing a sea change, and such an opportunity rarely arises." ”

More optimists believe that the move could be an "iPhone moment" for artificial intelligence, which could weaken or disrupt Google's dominance in the search market.

In October last year, Microsoft's Bing incorporated OpenAI's latest version of its image generation model, DALL-E3, to help visitors generate realistic images and increase user frequency.

According to Jordi Ribas, Microsoft's corporate vice president and head of engineering and product for Copilot and Bing, "With the addition of DALL-E3, Bing usage has increased by a factor of 10, which has really had an impact on product engagement and users. ”

In the second quarter of 2023, Bing monthly active users in the U.S. more than doubled year-over-year to 3.1 million, and the time users spent on the search engine increased by 84%, according to a Bloomberg analysis. SensorTower data shows that by the end of the year, Bing's monthly active users had grown steadily to 4.4 million.

However, it's still too early to talk about Bing challenging Google's supremacy. StatCounter data shows that Microsoft's Bing accounted for only 3.4% of the global search market at the end of 2023, and its market share has risen by less than 1 percentage point since the announcement of the integration of ChatGPT, while Google still accounts for 90% of the total global search market.

When it comes to search, data is a key factor in determining the quality of a search, i.e., the more searches, the more accurate the search answers.

How generative AI will change search and ranking results will also have an impact on the advertising model that Google relies on. In the past, search ads mainly relied on clicks to make money, resulting in a large number of inferior advertising messages in front of the front, and the user experience was extremely poor. In China, a similar situation exists.

"If we can build a better search engine than Google, then we should do it. "Google shows you 13 ads and 10 blue links, and maybe there's a better way to help people search for information and act on the prompts," Sam Altman noted. ”

However, in order to win the search market, it also needs the in-depth cooperation of hardware companies and content data parties. For now, OpenAI is still struggling with search data.

As recently as April 30, eight daily newspapers owned by Alden Global Capital sued OpenAI and Microsoft, accusing the two companies of illegally using news articles to power their AI chatbots.

In the complaint, the companies allege that OpenAI and Microsoft used millions of copyrighted articles to train and feed their generative AI products, including ChatGPT and Microsoft Copilot, without permission. The lawsuit did not seek specific monetary damages, but called for a jury trial and said that the publisher should be compensated for using the content.

"We've spent billions of dollars gathering information and reporting news in publications, and we can't allow OpenAI and Microsoft to expand and steal our work and build their own businesses at our expense. Frank Pine, executive editor of the Alden newspaper, said in a statement.

An OpenAI spokesperson responded that the company was not aware of Alden's concerns before, but was building partnerships and conversations with a number of news organizations to explore collaboration opportunities. On May 6, local time, the Financial Times announced that it had reached a content licensing agreement with OpenAI, which will be able to use its database to train GenAI models.

At the same time, more than 8,500 authors, including Margaret Atwood, Dan Brown and Jody Picourt, have joined forces urging tech companies such as ChatGPT and Bard responsible for generating AI applications to stop using their work or offer compensation without proper authorization.

A recent New York Times survey found that many tech companies are ignoring policies in their efforts to keep up with AI and are discussing circumventing copyright laws to get as much data as possible to train chatbots. Domestic large models can not avoid this problem, and the reference answers of various large model products mostly come from media reports or existing data.

Generative AI will fundamentally change the search market, and the day may not be too far off. But before they can do that, these AI companies need to first settle issues such as data sources and copyright disputes.