Estimated read time: 5-6 minutes
This archived news story is available only for your personal, non-commercial use. Information in the story may be outdated or superseded by additional information. Reading or replaying the story in its archived form does not constitute a republication of the story.
REDMOND, Washington — An entire generation of internet users has approached search engines the same way for decades: Enter a few words into a search box and wait for a page of relevant results to emerge. But that could change soon.
This week, the companies behind the two biggest U.S. search engines teased radical changes to the way their services operate, powered by new artificial intelligence technology that allows for more conversational and complex responses. In the process, however, the companies may test both the accuracy of these tools and the willingness of everyday users to embrace and find utility in a very different search experience.
On Tuesday, Microsoft announced a revamped Bing search engine using the abilities of ChatGPT, the viral AI tool created by OpenAI, a company in which Microsoft recently invested billions of dollars. Bing will not only provide a list of search results, but will also answer questions, chat with users and generate content in response to user queries.
The next day, Google, the dominant player in the market, held an event to detail how it plans to use similar AI technology to allow its search engine to offer more complex and conversational responses to queries, including providing bullet points ticking off the best times of year to see various constellations and also offering pros and cons for buying an electric vehicle. Chinese tech giant Baidu also said this week that it would be launching its own ChatGPT-style service, though it did not provide details on whether it will appear as a feature in its search engine.
'Our story has just begun'
The updates come as the success of OpenAI's ChatGPT, which can generate shockingly convincing essays and responses to user prompts, has sparked a wave of interest in AI chatbot tools. Multiple tech giants are now racing to deploy similar tools that could transform the way we draft emails, write essays and handle other tasks. But the most immediate impact may be on a foundational element of our internet experience: search.
"Although we are 25 years into search, I dare say that our story has just begun," said Prabhakar Raghavan, a senior vice president at Google, at the event Wednesday teasing the new AI features. "We have even more exciting, AI-enabled innovations in the works that will change the way people search, work and play. We're reinventing what it means to search and the best is yet to come."
For those who may not be sure what exactly to do with the new tools, the companies offered some examples, ranging from writing a rhyming poem to helping plan an itinerary for a trip.
Lian Jye Su, a research director at tech intelligence firm ABI Research, believes consumers and businesses would be happy to embrace a new way to search as long as "it is intuitive, removes more friction, and offers the path of least resistance — akin to the success of smart home voice assistants, like Alexa and Google Assistant."
But there is at least one wild card: how much users will be able to trust the AI-powered results.
A new search experience, and new problems
According to Google, Bard can be used to plan a friend's baby shower, compare two Oscar-nominated movies or get lunch ideas based on what's in your fridge. But the tool, which has yet to be released to the public, is already being called out for a factual error it made during a Google demo: It incorrectly stated that the James Webb Telescope took the first pictures of a planet outside of our solar system. A Google spokesperson said the error "highlights the importance of a rigorous testing process."
Bard and ChatGPT, which was released publicly in late November OpenAI, are built on large language models. These models are trained on vast troves of online data in order to generate compelling responses to user prompts. Experts warn these tools can be unreliable — spreading misinformation, making up responses and giving different answers to the same questions, or presenting sexist and racist biases.
There is clearly strong interest in this type of AI. The public version of ChatGPT attracted a million users in its first five days last fall and is estimated to have hit 100 million users since. But the trust factor may decide whether that interest will stay, according to Jason Wong, an analyst at market research firm Gartner.
"Consumers, and even business users, may have fun exploring the new Bing and Bard interfaces for a while, but as the novelty wears off and similar tools appear, then it really comes down to ease of access and accuracy and trust in the responses that will win out," he said.
Generative AI systems, which are algorithms that can create new content, are notoriously unreliable. Laura Edelson, a computer scientist and misinformation researcher at New York University, said, "there's a big difference between an AI sounding authoritative and it actually producing accurate results."
How to mimic authority
While general search optimizes for relevance, according to Edelson, large language models try to achieve a particular style in their response without regard to factual accuracy. "One of those styles is, 'I am a trustworthy, authoritative source,'" she said.
On a very basic level, she said, AI systems analyze which words are next to each other, determine how they get associated and identify the patterns that lead them to appear together. But much of the onus remains on the user to fact-check the answers, a process that could prove just as time-consuming for people as the current model of scrolling through links on a page — if not more so.
Microsoft and Google executives have acknowledged some of the potential issues with the new AI tools.
"We know we won't be able to answer every question every single time," said Yusuf Mehdi, Microsoft's vice president and consumer chief marketing officer. "We also know we'll make our share of mistakes, so we've added a quick feedback button at the top of every search, so you can give us feedback and we can learn."
Raghavan, at Google, also emphasized the importance of feedback from internal and external testing to make sure the tool "meets the high bar, our high bar for quality, safety, and groundedness, before we launch more broadly."
But even with the concerns, the companies are betting that these tools offer the answer to the future of search.
Contributing: Clare Duffy, Catherine Thorbecke and Brian Fung
