On Wednesday, Microsoft revealed a revised iteration of its search engine, Bing, with new features based on the same AI technology powering chatGPT, the large language model AI that has taken the world by storm since its launch in late 2022. The company is also bringing new AI-powered enhancements to its Edge browser, vowing that combining AI with search and browsing will usher in a new era for web surfing and online information discovery.

“It’s a new day in search. The race starts today, and we’re going to move and move fast,” said Satya Nadella, Microsoft’s CEO, introducing the new features. “Most importantly, we want to have a lot of fun innovating again in search because it’s high time.”

Microsoft showed some demo cases of the “new Bing”. During a search, the AI model can provide annotations or chat and interact with users similarly to ChatGPT. The most important difference between the model employed by Microsoft (GPT 3.5) and ChatGPT (GPT3) is that the AI-powered “BingGPT” can now collect information about current affairs and events and use it to reply to users’ queries. OpenAI’s publicly available chatbot, on the other hand, was trained with data collected before the summer of 2021. The new AI-powered Bing is already live for previews limited to desktop users who sign up on a waitlist.

Microsoft has historically trailed behind Google in Search, with Bing lacking the traction and Google Search’s critical mass of users. Yet, when integrating AI into the engine, Microsoft enjoys a healthy first-to-market advantage. The company invested early into Sam Altman’s OpenAI, the startup behind ChatGPT, and recently it announced a further round of financing for an undisclosed sum that industry sources place at about $10 billion. 

Pressured by Microsoft’s investment and new products, Google has recently felt the pressure to show results from its internal teams working on AI large language models (LLM) and to explain how it plans to counter the threat of LLMs taking over the web search market. 
Big G’s response came in the form of “Bard”, a new conversational AI that the company plans to integrate into Google Search relatively soon and that, for now, is only available to a small cohort of selected testers.

Google called Bard an “experimental conversational AI service” based on its GPT competitor, LaMDA. According to Google’s CEO, Sundar Pichai, “Bard can be an outlet for creativity, and a launchpad for curiosity, helping you to explain new discoveries from NASA’s James Webb Space Telescope to a 9-year-old, or learn more about the best strikers in football right now, and then get drills to build your skills.”

While it’s fair to assume that Google had been working on Bard for quite some time, the somewhat unexpected public release of ChatGPT last November forced the company to ready a customer-facing version of the product on very short notice. As a result, one of the demos Google used to show Bard’s functionalities contained a factual error, claiming that the James Webb Space Telescope took the first picture of an exoplanet (which is false).

The blunder is proof of a worrying tendency of LLMs to present their own “hallucination” as truths. But while ChatGPT’s mistakes, widely remarked on social media, didn’t make it to Microsoft’s presentation, Google’s public mishap was seen as a sign of the company’s alleged unpreparedness to face Microsoft’s competition. As a result, Wall Street quickly shaved a hefty 7% off of Alphabet’s shares, costing Google’s parent company about $100 billion. 

While eager to show what AI can do to change search, Microsoft and Google have yet to answer a series of fundamental questions about integrating LLMs into their search engines. Namely: if the future of search will go through conversational AIs that will provide users with answers without ever leaving the search page, what will happen to the economy of the Web?

Moreover, if Bing and Google Search will stop sending users and visits from their search engines to the websites they source their information from, how are these websites supposed to keep getting published? And as a final consequence, if publishing a website won’t be a viable business in the future, where are Bing and Google Search expecting to fetch the information they need to present results and actually train their Large Language Models?

Since neither Google nor Microsoft provided comprehensive answers to these questions, we put them to chatGPT. Here is what AI has to say in its own defence: 

These are important questions that address the implications of integrating AI into search engines. The future of web search could indeed be affected by the use of conversational AI that provide answers directly to their users, but this could also impact the economics of the web.

If search engines stop sending users and visits to the websites where the information comes from, these sites may no longer be able to continue to be published and the publishing business may become less profitable. However, it is important to note that this scenario may not be likely or, if it happens, may not happen quickly.

Regarding the information that Bing and Google Search would use to present their results and train their language models, these search engines might continue to use different sources of information, such as public databases, information provided by companies, and online information sources. It might also be possible that a different economic model would develop that would allow websites to be published and monetised in a different way.