The dawn of personalised Web pages is powered by AI search
Readers First Initiative Blog | 26 February 2024
New AI-powered browsers and search engines flip the script on how we find and consume online content. It’s not just about links anymore, nor AI summaries. Arc Search creates bespoke Web sites based on user queries.
Arc Search is an iPhone browser with an embedded search engine developed by The Browser Company in the United States.
This morning I dictated: “What’s news in the U.K.?” and touched the button “Browse for me.”
Arc Search scanned BBC, Sky, The Guardian, and some other Web sites.
It showed a personalised Web page that summarised the headlines and down the page — the key points of the top stories.
The feature “Browse for me” has become an early users’ favourite: “epic,” “brought back the magic into search,” “future of browsing.”
Why it matters
This leap signifies a move towards highly customised experience, tailored not just to what you're looking for but also how you like it presented.
Many tech companies innovate with generative search experiences:
Microsoft Copilot, Perplexity AI, and others browse multiple sites and summarise results, but their interface is more similar to a chat than browsing the Web.
Microsoft Bing user also needs to click for a Copilot’s answer; in default, she gets a traditional list of links.
Google is experimenting with combining the two — an AI-generated summary and links — although it patented last fall a bolder experience: the summary is served by default, and the user needs to click to expand the links.
At the premiere of his new chatbot Gemini, Google CEO Sundar Pichai told Wired that he wanted to push “the boundaries of what’s possible.”
What will Google do
What he decides matters most, as Google enjoyed 91% share in the search market worldwide in January, per StatCounter. Microsoft’s Bing had 3% share. Startups like Perplexity and Arc Search hid in the Other browsers’ share of 0.05%.
Search is a top referrer for online traffic to news Web sites worldwide. According to Chartbeat, depending on the world’s region, between 17% and 41% of all pageviews on news sites in 4Q 2023 came from search — and primarily from Google.
Google estimated an AI-generated answer cost 10 times more than a standard keyword search, and that was slowing down the release. But the costs are falling, and the user demand and competitors’ innovations pressure the leader.
Sending less traffic to Web sites might put at risk US$31 billion in revenue that Google made in 2023 from its ad network on publishers’ Web sites.
But Google might wish to prioritise protecting US$175 billion that it made from its search engine, per company reports.
The chain reaction
Last June, I estimated the risks for publishers from the AI search revolution, expecting a couple of years for these risks to materialise due to high costs, concerns around accuracy of AI answers, and unknown demand.
A year later, concerns around accuracy haven’t disappeared, but surveys showed that consumers wanted AI search. Tech companies noticed. Per the Information, after observing how people used ChatGPT, OpenAI decided to develop its own Web search product.
Here’s how a chain reaction unfolds across the digital ecosystem, with cascading effects on all stakeholders.
Consumers want personalised content, and most get their news through aggregators already. Thanks to advances in Generative AI, the vision of personal media is finally technically feasible.
This inspires innovation in the value chain — device manufacturers, operating system providers, Web browsers, search engines, and AI developers.
Many tech companies already are vertically integrated, e.g., Apple, Google, and Microsoft do all or most of the jobs in the value chain. They can deliver what consumers want.
More integration is likely: AI developers and Web browsers want to become search engines (OpenAI, Arc Search) so they can differentiate and disrupt slower or less innovative incumbents.
All these companies aggregate audiences and content. For them, publishers are complements, and most online content is a commodity.
Implications for the media industry
The tech companies have enough market power to impose rules on publishers today. AI is making the Big Tech’s grip on the media even stronger, wrote Oxford researchers Felix Simon and Luisa Isaza-Ibarra in a new report.
The AI and search companies can sign licensing deals with selected publishers, but they don’t need deals with all of them. Some publishers — e.g., branded media, public, state-owned, or non-profit — might not mind getting scrapped or summarised, as it’s irrelevant to their funding model.
Decreased direct traffic and ad revenue, alongside the pressure on subscription funnels, necessitate a strategic rethink of publishers that need a sustainable revenue model to fund themselves.
Publishers might consider defensive actions like blocking AI and search bots, or pivoting to closed platforms and proprietary apps to protect their work and revenue streams.
They might also fight for their copyrights in courts or ask regulators and lawmakers for protection against the market failure and abuse, as Daily Maverick CEO Styli Charalambous suggested in his opinion piece for INMA.
Greg’s Readers First newsletter is a public face of a revenue and media subscriptions initiative by INMA, outlined here. INMA members may subscribe here.