AI search is about to transform how we consume online information. Some of these effects will be immediate and others a little farther out on the horizon. What is clear is that search is fundamentally going to change.
AI Search Will be a Game-Changer. Here's Why
At present, Google is starting to look more and more like a haystack, with us having to sift through sponsored ads and subpar intent matches, to find the needle. With LLMs, the answer will be laid out directly before you. The intent and the result will be an almost perfect match. The only one doing the searching will be the LLM.
There’s an additional feature of AI Search that will likely lead to a siphoning from traditional search: the ability to ask follow-up questions to refine your query. This feature will help you fine tune your searches, all almost perfectly aligning search intent with output–something that the Google algorithm doesn’t always get right. In this case, the intent is contextually based, taking all the little nuances of your conversation into account. With Google, too many specifics overwhelm the algorithm. In many ways, the intent ceiling is limited by traditional search in a way it is not with LLM-based search.
AI Search Agents: What You Need to Know
Slightly farther off, but perhaps no more than a couple of years, are AI agents. Think of this as an LLM with a highly specific task to do. Now think of many of these LLMs–or agents–”communicating” with each other. For example, think of a blog post in which one LLM does the writing, another does the reviewing, and another does the editing. Currently, this has to hard coded by engineers, but essentially you’d press a button (or enter a prompt) and that agentic workflow would play out.
In the coming year and beyond, agents are likely to play a huge role in how we engage with the web across the board, search being one of these areas. Here, agents, in collaboration with other agents, will be able to do many of the things we currently do: a Google search, clicking on relevant links, finding the information you want, vetting it against multiple sources, when necessary, and other sundry substeps. AI agents are already doing behind the scenes work in certain industries; coupled with Claude 3.5’s recent ability to control your mouse, AI agents accessing your computer will hardly be the thing of science fiction.
Soon, we will need little more than our voices to activate an agentic network. For example, we can ask them to plan a vacation for a family of four to Disneyland on a fixed budget. The AI agent will go to multiple AI search tools and find this information. Then, it’ll correspond with another agent to deliver a few options with a detailed budget breakdown in Excel. In this world, with agents acting as intermediaries between your question and the answer, we will be essentially removed from the whole search process.
As AI-driven agents streamline search, we’re also likely to see an impact on the financial structure of the internet as we know it. With fewer touchpoints between the internet and a user, there’ll be fewer opportunities for monetization. This will certainly impact revenue related to ads, unless LLMs decide to monetize their magic by popping in 15-second ads perfectly peddled to your predilection—something that is certainly within the realm of possibility. This agent-driven AI search will play out over years, with people trying to figure out how to generate revenue for targeted information within the walled garden of the LLM.
Expect Immediate Impacts on SEO!
What is going to be more immediately affected is Google SEO. One thing we might see is fewer people reflexively relying on Google. While AI Searches account for only a sliver of overall search traffic–exact numbers are hard to pin down, so nascent this field is–the sheer number of users of ChatGPT will inevitably begin using Search. Granted, AI has to contend with hallucinations that will make some users wary of completely relying on AI search. But given the near perfect intent alignment, users might come to find the off chance hallucination palatable. What this could translate to is AI search claiming an ever growing share of the overall search market.
Even if Google adapts–as it likely is already–by offering their own AI search within an LLM interface, Google potentially will play a part in rendering the current SEO paradigm obsolete. To be sure, people will always use Google search—after all, that is what they are used to. And many will continue to be wary of AI in a way that they are not wary of Google search, whether because of hallucinations or a general guardeness to anything AI-related. But even if AI search can account for 15% of all searches, the traditional paradigm is going to be shaken.
But the search titan itself is not going to be the only one impacted—far from it. In fact, most businesses are going to be affected to the extent that they rely on SEO traffic to generate their sales. At least from what I’ve been able to glean so far from playing around with GPT search and Perplexity, what an LLM recommends is not always the same as what Google search does.
Searches Will Feel Very Different!
We might also interact differently with LLM search outputs. A list of 5 of the top televisions to buy along your exact specifications is a lot easier to navigate than a long list of search results buried amongst ads, answers, and trending news. In other words, whereas we might click only on the first two links, choosing whatever is recommended there, with LLMs it’s easier to parse the range of choices. Suddenly, the two businesses that land the top Google hits don’t control the lay of land, at least as much.
Even if we interact with the outputs differently, you might still be wondering: how do we know that LLM search outputs will differ from the Google search results? This is a great question and something I’ve been researching a bit myself, using both Google and GPT Search/Perplexity. AI search prioritizes content that is contextually rich. Content that connotes a sense of authority and expertise will be valued by the AI more than content that follows the header structure the Google algorithm tends to prize.
For example, AI search results on “best TVs of 2024” would give more weight to the tech writer with a loyal following but whose posts haven’t had a high-touch SEO treatment than to a big name electronic provider whose posts are perfectly optimized but short on content specifics and general user engagement.
This discrepancy is notable because it provides another means by which AI search might gain on Google search (assuming Google doesn’t change its algorithm to match what AI does.)
The Takeaway
As these changes take hold, both content creators and consumers will need to adapt. With AI setting new standards for relevancy and authority, content creators may need to rethink how they structure and deliver information. In other words, get ready to rethink those algorithms—AI search is here.