It’s hard to imagine the internet without Google.
The tech giant has become so synonymous with searching the web that it has become a verb – we don’t look it up, we “Google it”.
Google ended 2022 as it ends every year, as the most visited website in the world. Its estimated share of the search engine market stands at 92% (Microsoft‘s Bing is its closest rival, on 3%).
On the surface, it might not like a landscape that’s ripe for change but don’t be so sure.
The launch of ChatGPT, an AI chatbot, last year threatened to upend how people prepare for job interviews, journalists write stories, and children do homework.
Trained on a huge amount of text from across the internet, with the ability to provide human-like responses to almost any prompt, it sparked speculation it could pose a threat to Google.
Search engines ready to bet big on AI
The New York Times reports Google founders, Larry Page and Sergei Brin, have been brought back to help add ChatGPT-like features to the search engine they launched more than 25 years ago.
Google boss Sundar Pichai reportedly wants to speed up the firm’s plans for conversational AI in its products and services, which go far beyond a chatbot that pretends to be a tennis ball (seriously, we tried it and it was really odd).
It comes as Microsoft makes a multibillion dollar investment in ChatGPT creator OpenAI, raising the possibility that it could find its way into products like Office (welcome back, Clippy and friends?) and – yes – Bing.
The potential AI arms race is one predicted by former Google advertising tsar Sridhar Ramaswamy, who wants to use what he learned during 15 years at the company, to get ahead of this potentially game-changing trend.
“We are at an interesting juncture,” he tells Sky News from his home in California, from where he co-founded startup search engine Neeva.
“Where large language models and AI offer unprecedented ability to peer into information, to sift through things and deliver answers in a way that simply was not possible before. This is a really exciting time for search – and I think it’s going to be disrupted in multiple ways by multiple companies.”
Stream Type LIVESeek to live, currently behind live
Remaining Time 0:00FullscreenCan machines have feelings?
How AI could change how you search the web
Among those would-be disruptors are the likes of You.com, a search engine launched out of California (where else) back in 2021, which added a bot called YouChat in December; and Neeva’s own new AI.
Neeva, which launched in the UK in October, aims to provide informative and reliable search results without being driven by user data and advertising.
Its AI functionality is being added for UK users in February. The AI trawls the web for information, produces a single, written answer to the query, and – like YouChat – cites each of its sources for users to find out more.
And it works in real time, meaning it stays up to date with current affairs and provides references accordingly for its admittedly small pool of one million monthly users.
“It gives search the power to be a lot more fluid to what it’s been so far,” says Ramaswarmy, who believes the advertising model he led at Google – which accounts for the majority of its revenue – needs to be challenged.
“The entire search experience becomes focused on just getting you to click on an ad,” he says.
“But there is a bigger reason – the obsession with ads on the internet has also steered the Google algorithm in a way that focuses on engagement, so more and more you see ‘made for Google’ sites that game how to get on top.”
But is it all upside?
Neeva’s commitment to being ad-free comes at a premium – £5.49 a month, or £44.99 a year.
The majority of its 600,000 users are on its free tier, which limits them to 50 searches per month.
Allowing limitless searches via powerful AI isn’t a particularly cost-effective business model, with OpenAI CEO Sam Altman admitting the computing costs to run ChatGPT are “eye-watering”.
Dr Andrew Rogoyski, from the Institute for People-Centred AI at the University of Surrey, says the infrastructure required to run such a service at the scale of a Google would be enormous.
“There is some way to go in streamlining conversational AI in a way that we can afford,” he tells Sky News.
“AI is getting bigger and consuming more energy, and that’s the wrong direction – it’s pushing it into the hands of big organisations.”
But there are more fundamental challenges that any search engine seeking to leverage AI will need to consider.
‘Incorrect or nonsensical’
Like ChatGPT, NeevaAI is a large language model, meaning it is trained on huge amounts of information.
But OpenAI acknowledges its answers can be “overly verbose” and “incorrect or nonsensical”.
“It doesn’t know right from wrong, it doesn’t know authoritative from gossip,” says Ramaswarmy.
Of course, it’s one thing for a nascent ChatGPT to make mistakes, another entirely if a firm like Google rolls out a commercial product with similar failings. As Microsoft found out after its 2016 chatbot was taught to say offensive things.
“Conversational AI is very believable, certainly in short bursts, and that will improve over time, but because it’s so believable and plausible, people will not necessarily challenge it,” Rogoyski says.
TOP RISKS IN USING AI FOR SEARCH ENGINES (ACCORDING TO CHATGPT)
- Bias: AI-powered search engines can perpetuate and even amplify existing biases, particularly if the data used to train the model is biased
- Privacy: AI-powered search engines can collect and store large amounts of personal data, which can be used for targeted advertising or other purposes
- Censorship: AI-powered search engines may be used to censor or suppress certain types of information
- Misinformation: AI-powered search engines may return misinformation or fake news, particularly if the AI model is trained on unreliable sources
- Job loss: AI-powered search engines may lead to job loss for certain types of jobs, such as librarians or research assistants
- Security: AI-powered search engines may be vulnerable to hacking or other types of cyber attacks, which could compromise user data or disrupt search results
- Dependence: People may become too dependent on AI-powered search engines, which could lead to a lack of critical thinking and research skills
- Monopoly: AI-powered search engines may lead to the creation of powerful monopolies, which could lead to a lack of competition and innovation
Are we really ready for change?
Seeing the potential for a shake-up on the internet is one thing, but seeing it through is another entirely.
Elon Musk’s tumultuous takeover of Twitter hasn’t upended people’s habits to the extent experts predicted, with the rapid rise in users at would-be rival Mastodon having stalled.
Ramaswarmy admits causing an internet “mass movement” is difficult, but views ChatGPT’s breakthrough as evidence that “a platform shift” is on the cards.
“Think of how we saw Microsoft, Nokia and Blackberry disappear from the mobile world and let Google and Apple become the dominant players,” he says.
“It feels like this is one of these face-off moments.”