A subtle way that artificial intelligence (AI) is inserting itself into everyday life is in the effect it is having on web browsing, social media and streaming content like movies. Tim Berners Lee invented the World Wide Web at CERN in 1989, and websites for the general public emerged around 1993. The issue was how to find them, and a series of competing web browsers emerged, including Netscape and Internet Explorer, as well as Mosaic, Opera and Ask Jeeves. A small company called Google was quite late on the scene, appearing in 1998. I can recall a colleague at work showing me for the first time Google Search’s stripped-down, simple interface that resulted in extremely quick and comprehensive results compared to other browsers. Everyone at the office switched to Google in short order since its result were so much better. Despite competition from the old guard of browsers and newer ones like Firefox and Safari, Google’s market share climbed, with Google Chrome growing from 3% market share in mid 2009 to 38% by early 2013, overtaking Internet Explorer. By mid 2016, Chrome was at 62% market share, a dominant position in the desktop browser market that it has never relinquished.
This comfortable position at the top of the pile has become challenged recently, not by a new browser but by AI. ChatGPT’s conversational interface allows users to ask questions in much the same way that they would use a web browser. The difference is that ChatGPT and its competitors can quickly scan numerous websites and bring back aggregated answers, meaning that users often don’t need to click through to the source websites. Google fought back by providing AI summaries of its own, but this has had unintended, and significant, consequences. The era of the “zero click search”, where users just rely on AI summaries, has arrived.
A December 2024 report by Bain found that 80% of consumers now rely on the AI summary answer in a web browser, and never click through to another site in at least 40% of their searches. This is a major problem for marketers, who have spent years on web advertising and search engine optimisation for their brands. Now, many consumers simply never see a final website or an advert attached to it. By mid-2025, there were around 80% fewer click-throughs to original content, as consumers see the AI summary produced by Google, or simply use a chatbot instead of a browser. The AI summaries provided by Google have a chilling effect on referral traffic. One study found that users only clicked a link under an AI summary one in a hundred times. The effect is dramatic: the popular news website MailOnline found a 56% drop in click-throughs. Advertisers now need to consider how to optimise for AIs rather than search engines, such as adapting content for semantic search. Videos and interactive content are a way to distinguish brands, but these require more effort to produce. There is a further troubling effect: given current AI’s tendencies to hallucinate, these AI answers may not always be correct.
For social media, AI has also had an impact. Sites like Facebook, YouTube, X (formerly Twitter), Instagram, WhatsApp and TikTok rely on AI to curate content tailored to each user’s preferences. Algorithms analyse the behaviour of users to decide which posts, adverts and videos to show to an individual. This has created the echo chamber phenomenon, whereby users are pushed content from like-minded people and sources, often resulting in a spiral into more and more extreme content niches. The platforms also use AI for content moderation such as screening out offensive posts. This is something that is expensive and difficult to do manually, given the sheer volume of user-generated content now being created. However, AI also brings new challenges in the form of deepfakes. A passable audio deepfake can now be produced from as little as three seconds of a recorded human voice, and if longer recordings are available (say, fifteen minutes) then the deepfakes are disturbingly accurate. This creates a minefield of potential disinformation, such as with political content, and potential fraud. Social media companies and governments have yet to effectively react to the newly improved deepfakes, which have the potential to cause chaos in political campaigns and to unleash rampant fraud. You may be able to spot that an unexpected email from a Nigerian prince offering you millions is probably a scam. On the other hand, it is trickier when you get a phone call or even video call from someone who appears to be your finance director or a family member asking you to transfer money urgently.
Streaming web content has long been an adopter of AI. Companies like Netflix use elaborate machine learning models to analyse viewing habits and preferences, and suggest what to watch next at the end of a show in line with what will most appeal to a specific customer, and so keep them engaged longer with the platform rather than switching channels. The mechanics of streaming a video itself also use AI, with adaptive streaming that optimises video content based on the throughput of the network in a particular geography, optimising quality for different devices. The technology allows augmented reality, where for example, a customer might be able to see sporting statistics overlaid for a live event, request replays or even see exclusive fan content for their team. Location-based marketing is possible, rather than just broadcasting the same advert to everyone watching a show. The depth of knowledge of consumer behaviour reinforces long-standing concerns about privacy, or the lack of it.
The penetration of AI into web browsing, social media and streaming changes how users discover content and how platforms engage with their audiences. Businesses, content creators, and consumers all need to be aware of the changes that AI is bringing to these familiar platforms, and businesses in particular need to adapt to the wave of change that is washing over the way that we all interact with these platforms.
