Googlers were shocked to learn in March that South Korean consumer electronics giant Samsung was considering replacing Google with Microsoft’s Bing as the default search engine on its devices, reports The New York Times.

Google’s reaction to the threat from Samsung was “panic,” according to internal reports seen by the publication. According to estimates, the contract with Samsung put $3 billion in annual revenue at stake. This added pressure, because a similar contract with Apple for $20 billion is also due to be extended later this year.

Artificial intelligence competitors like the new Bing are fast becoming the most serious threat to Google’s search business in 25 years, and in response the company is rushing to build an entirely new search engine based on the technology. According to internal documents seen by the publication, the company is also modernizing the existing search system by adding artificial intelligence functions to it.

New features, called Magi, are created by designers, engineers and executives who work in so-called sprint rooms to tweak and test the latest versions. The new search engine will offer users a much more personalized experience than Google’s current search by trying to anticipate user needs.

Google has been concerned about AI competitors since OpenAI, a San Francisco-based startup that works with Microsoft, showed off a chatbot called ChatGPT in November. About two weeks later, Google created a task force within its search division to begin building artificial intelligence products, two people with knowledge of the effort said, but were not authorized to discuss it publicly.

The threat from Samsung is the first potential crack in Google’s seemingly impregnable search business, which was worth $162 billion last year. While it wasn’t clear whether Microsoft’s AI work was the main reason Samsung was considering a change after 12 years, Google suggests it was. The contract is under negotiation and Samsung could still stay with Google.

But the idea that Samsung, which makes hundreds of millions of smartphones running Google’s Android software each year, is even considering changing its search engine has shocked Google employees.

Google has been doing research in the field of artificial intelligence for many years. Its DeepMind lab in London is considered one of the best AI research centers in the world, and the company has pioneered AI projects such as self-driving cars and so-called large language models used to develop chatbots. In recent years, Google has used large language models to improve the quality of search results, but has refrained from fully implementing AI because it has a tendency to generate false and biased claims.

The priority now is to gain control of the next breakthrough technology in the industry. Because of this, last month Google departed from its own principles and released the Bard chatbot, which received mixed reviews.

Plans for the new search engine, which show Google’s ambitions to reimagine the search experience, are still in the early stages, and there is no clear timeline for when the company will release the new search technology.

The system will learn what users want to know based on what they are searching for once they start using it. And will offer lists of pre-selected products to buy, useful information and other data. It will also be more dialogic, like a chat with a specialist.

But long before the search engine is rebuilt, Project Magi will add features to the existing search engine, according to internal documents. More than 160 Google employees are working on it, according to a person familiar with the matter.

Magi will keep ads in the mix of search results. Search queries that could lead to a financial transaction, such as buying shoes or booking flights, will still have ads on the results pages.

This is important for Google because search advertising is the main way of making money. The Bard chatbot is currently ad-free, and there are expectations in the tech industry that AI responses to search engine queries could make ads less relevant to users.

Scheduled search additions can also answer software coding questions and write code based on a user request. According to the document, Google can serve ads under these answers as well.

Last week, Google invited some employees to test Ma’s featuresgi and asked them to ask the search engine additional questions to assess its conversational ability. Google is expected to release the tools to the public next month and add more features in the fall.

The company plans to initially release these features to a maximum of one million people. By the end of the year, this number should gradually increase to 30 million. Features will be available exclusively in the United States.

Google is also considering allowing people to use Google Earth’s artificial intelligence mapping technology and search for music by talking to a chatbot, a Google executive wrote in the document.

Other ideas are at various stages of development. The tool, called GIFI, will use AI to generate images in Google Image search results. Another tool, Tivoli Tutor, will teach users a new language through open text dialogues with artificial intelligence.

Another product, Searchalong, will allow users to ask questions of a chatbot while surfing the web in Google Chrome. For example, people can ask a chatbot about activity near an Airbnb rental, and the AI will scan the page and the rest of the web for an answer.