Recently, AOL discontinued the dial-up tone synonymous with the internet. After 34 years, we will no longer hear the mechanical screech of connecting to the World Wide Web. Nostalgia notwithstanding, this news gives me pause to contemplate what my children will see sunset in their lifetimes.
Consider Google today. A tool so permeated into our modern society that it has become a verb (Google). While the creators may not have set out to lasso the entirety of human knowledge, rivaling that of the mythical Library of Alexandria, Atlantis, or the Great Sphinx, perhaps they did hold such a dream when naming the thing (googol is a 1 with 100 zeros behind it).
And as the saying goes, “It’s hard to argue with a result.”
Even more difficult, it’s hard to imagine such a ubiquitous tool could become obsolete when nearly 70% of the world’s population has access to the internet. However, we’ve started to see the cracks in the foundations of traditional web search. What will replace it?
The answer: AI.
What is zero-click search?
A zero-click search is one where the user finds what they need without having to click a single link. Generative AI, powered by large language models that summarize information instantly, now delivers those answers directly in the search results. Forbes has labeled zero-click search as “the 60% problem,” referring to the decline in website clicks that occurred due to AI overviews.
From this “click-less” reality, we derive the name of the very same phenomena that will change the popular dynamic by which we seek information. The AI overview is made possible by generative AI, which is based on large language models (LLMs) that can learn from text, videos, or code. Scouring websites for ungated information, LLMs can summarize in fractions of seconds without users ever clicking through to a website.
The troubling reality is this is a self-cannibalizing feature for traditional search…
Advertising revenue and SEO add value by generating web traffic – often measured by clicks, time on page or lead generation. Conversion rates for this channel tend to be small, fractions of a percent, so spend becomes important, especially if you want to show up on “page 1.” Entire departments are dedicated to promoting content that will “show up” to interested parties and potential customers.
So, what happens when clicks decline along with results? Will CMOs continue to spend millions of dollars out of their marketing budgets to pay for, well, zero? Of course not. They will shift their strategy and spend.
Interestingly enough, Google Ad results, while positive in Q4 2024 (growing over 10%), missed analyst expectations. Rather than signaling decline, this shift underscores how even the most powerful platforms are redefining themselves for an AI-first future. AI overviews are only the beginning. Along with Google, Microsoft, OpenAI and other major players are actively developing agentic AI systems.
We see Google investing $24 billion in AI (including data centers). In its own product news, Google touts the impact of its AI overviews and its LLMs (Gemini) capabilities, driving a 10% increase in usage. The same article goes on to outline deeper capabilities like Google Lens and the ability to make purchases using Google Pay.
This flips the value proposition of the world’s most popular search engine – from gathering information to taking action.
So what happens next?
The prediction
By 2027, traditional web search will be obsolete. The quality, power and veracity of the hyperscalers' AI will deliver an experience that no longer requires people to click on websites. Tasks from the pedestrian to the complex will be completed through seamless interactions with LLMs.
The company synonymous with the concept of search has already laid the foundation to replace its most robust income stream with AI capabilities that will forever reshape how we live, work, relate, create (destroy) and discover.
In 2023, Bill Gates penned a prediction that “everyone will have an AI-powered assistant within five years (employed or not).” This sentiment, while accurate, is ill-timed – because it’s happening now. And for that proof, we only have to look at the meteoric rise of OpenAI and ChatGPT.
Launched in November 2022, ChatGPT brought AI into our everyday lexicon. Beforehand, outside of Silicon Valley, only the tech elite knew the name Sam Altman. And in under three years, OpenAI has now hit $10 billion in annual recurring revenue, driven by subscriptions. Wildly, there are only about three million paying business users, dwarfed by the 500 million weekly active users. While concern from Wall Street continues to compare today’s AI economy to a bubble akin to the late 1990s dot-com debacle, the results speak for themselves. People are paying $20 a month (sometimes more) to gain access to AI that allows them to work faster, plan better and live easier.
So, what does ChatGPT have to do with the downfall of web search?
The answer to that question: agentic AI.
It’s agentic
No, not electric (boogie, woogie, woogie), agentic!
SAS defines agentic AI as including five key components: perception, cognition, decisioning, action and learning. While definitions and standards vary, it is generally agreed that an “agent” has a level of autonomy to perform the work on your behalf (much like an assistant). They are not all-powerful, but will pursue their goals with tenacity.
AI Agents: What they are and why they matter
It's this pursuit that will make agents popular with people.
Like water, humans take the path of least resistance. At least one study from 2021 finds, we are ‘hard-wired’ to do this. In fact, at this year’s All Things AI conference in Durham, NC, Sriram Raghavan of IBM stated unequivocally, “I do not want artificial general intelligence (AGI), I want artificial useful intelligence.”
If an AI agent can, with the same natural language processing that enables you to talk to Siri, buy an insurance policy, order groceries, reserve a plane ticket, send your autonomous vehicle via remote to the repair shop to get an oil change… Wouldn’t you spend a nominal amount of money to “hire” that agent?
Why Google? Ask the agent. And get it done.
At SAS, we’ve already built an agentic AI solution to solve internal knowledge management searches, called SAS® Retrieval Agent Manager (RAM). This is just the first step.
As agentic capabilities mature beyond narrow use cases and become popularized by hyperscalers, people will yield their autonomy more and more – just as we did when we stopped going to encyclopedias for answers.
Does web search go away? I think so. Maybe even Google becomes a purveyor of agents. Stranger things have happened, until people thought to do it (I still can’t believe we carried around luggage before we thought to put wheels on bags).
