The way the world uses the internet will change forever with the rise of artificial intelligence-powered search engines, which experts predict could throw up a host of issues in an already fraught field.
The technology has the potential to disrupt or even cripple business models, while for users there is another layer of information to scrutinise when people are already at times drowning in it.
Microsoft and the Goliath of search, Google, have been locked in an arms race to get more powerful AI into their search engines.
While the shift isn’t expected to lead to an overhaul of the underlying algorithms that find content, it will affect the material users see from their searches – for example, leading off a list of links with a summary of the top results.
Microsoft recently revealed it was integrating a “next-generation” model of OpenAI into its search engine Bing.
The “new” Bing generates a list of traditional search results and an AI answer beside it.
A demo site gives the example of a suggested three-course menu, kicking off with a starter of wild mushroom tartlets or vegan popcorn tofu nuggets.
There is a waitlist to preview the site in action and, so far, more than a million people have been granted access.
Microsoft this week announced the preview was expanding to mobile devices.
The company is yet to reveal a date for when the new Bing will go live to the wider public.
Google, meanwhile, has announced an experimental conversational AI service, Bard, which is open to a group of select “trusted testers” and will be more widely available in the coming weeks before the technology is integrated into its search engine.
Then, there’s the smaller players like AI-powered search engine Neeva, launched in Australia by ex-Google Ads vice president Sridhar Ramaswamy.
The platform spits out an AI-generated answer in response to users’ queries at the top of a list of traditional search results. This answer, like with the new Bing, includes citations with links to websites.
“Early last year, we realised that AI could be a game changer for a variety of reasons, the biggest one being that all of a sudden, we had language models that were capable of performing pretty complicated tasks, like writing a short four-sentence summary of a 1500-word blog,” Mr Ramaswamy said.
“What we do is we generate short 80- to 100-word answers with citations, with real-time information, that gets to the heart of what it is that a person is looking for”.
Neeva runs off a subscription-based model, charging Australian users $7.99 a month to access all its search and personalisation features. It also offers a free basic service.
The search engine is notably ad-free – a big departure from Mr Ramaswamy’s previous role with rival Google.
“Our biggest practical issue has been getting distribution,” he said.
“But you show these experiences to any person on the street, and they instinctively go, ‘that’s amazing’.”
AI search engines have the potential to give users a much richer experience, experts say.
However, they create a big problem for advertisers and businesses with the possibility that people will take the first answer at face value and not bother clicking through to other sites.
The potential AI-related issues come on top of existing criticisms that search engine results create “filter bubbles”, exerting undue influence on the information people see.
Advertisers already dominate Google’s first few search results and the revenue-raising challenge will be amplified with artificial intelligence-powered search, University of New South Wales AI Institute chief scientist Toby Walsh suggests.
“We’re not going to be clicking so often because the answer is actually already coming back to us,” he said.
“Then how are they going to raise the revenue? Are they going to collect information on us from the queries we type and sell that to advertisers?”
The integration of AI into search engines poses an “existential threat” to a company like Google given the revenue issue and the possibility users could go elsewhere, according to Prof Walsh.
Ultimately, one of the big tech players will win the search engine race.
An issue also arises for publishers, with internet users less inclined to check a source and ongoing arguments about media bargaining and fair use likely to be amplified, Prof Walsh added.
However, a more fundamental problem is that chat bots sometimes make stuff up.
“They’re not really understanding, they’re just saying what’s probable,” Prof Walsh said.
“If people aren’t following the links, checking for themselves, then truth is going to be an even more fungible idea than it is today.”
Bard has already run into ridicule thanks to a misstep when it made a factual error about NASA’s James Webb Space Telescope, a kink Google says highlights the importance of its “rigorous testing process”.
For Monash University dean of information technology Ann Nicholson, AI search engines will add another layer that people will need to be wary of when they’re taking in information.
In turn, this meant educators would need to adapt their lessons to show students how to use tools and about their limitations, she said.
If companies get AI search engines right, they will be taken up widely and prove disruptive, Prof Nicholson added.
However, she describes it as a “step change” for technology and cautions society can’t yet be sure how it will play out.
For his part, Mr Ramaswamy believes this is the “iPhone moment” for search.
Prof Walsh agreed with the sentiment, predicting AI-powered search engines could overtake traditional models in the next year or two.
“When the iPhone came out in 2007, the first iPhone was really pretty clunky,” he said.
“You look at what’s happened in the 15 years since and the technology has become much more sophisticated, smoother and reliable.
“We’re only beginning that journey here today.”
By Cassandra Morgan in Melbourne
