What is AI, actually, and how is it affecting SEO?

If it feels like the entire internet woke up one day and decided to start every sentence with “AI,” you’re not wrong.

Marketers are being hit with a daily wave of LinkedIn thought leaders, half-baked prompt hacks, and promises that ChatGPT is either going to 10x your productivity or take your job entirely.

And in the middle of all this? You (the digital marketer).

Marketers are trying to figure out if this is just another buzzword cycle or the beginning of a complete rewrite of how we do content, SEO, PPC, reporting, and, well, everything.

So let’s break it down. 

Consider this your AI starting guide, written for marketers who are tired of needing a younger person to translate all the jargon the way we once had to help our parents get the internet working or open an AOL chat window back in the day.

Defining AI and LLMs (and why they matter)

I promise I’m not asking “what is AI” to hit a perfect keyword density. I want to establish a common ground so that we’re all aligned moving forward on what I refer to as “AI.” 

You’ll see below that many of these terms are used interchangeably and even inappropriately. You want to sound smart in front of your friends, right? Right!  

At its core, artificial intelligence refers to machines performing tasks that typically require human intelligence: understanding language, recognizing patterns, making decisions, and yes, even generating content.

The type of AI that is making everyone excited right now is generative AI—models that can produce text, images, code, and more, based on patterns they’ve learned from enormous datasets. 

Tools like ChatGPT, Gemini, and Claude don’t “think” the way humans do—they predict the next most likely word or phrase based on what they’ve been trained on.

Despite what people want it to be, AI is not a silver bullet solution to getting rich quickly. It’s not going to automate everything overnight or instantly reduce headcount to make that profit and loss look even better. 

It’s data aggregation, at scale. Large Language Models (LLMs) aren’t producing net new data. They are simply processing the data across the web to provide a solution strictly based on internet consensus. 

Generative AI and LLMs represent a massive shift for content creators and marketers. You’re no longer just optimizing for a classic search engine click—the goal now is to create content that can be effectively interpreted and summarized by machines. 

The critical implication is the rise of zero-click search results: AI systems can summarize, cite, and present your content directly to users, often answering their queries without requiring them to visit your website. (Examples include Google’s AI Overviews and ChatGPT answers.)

This fundamentally shifts the SEO landscape away from a traffic game to an authority and data-ingestion game.

That’s why understanding how AI works and what it doesn’t do well is critical before deciding where it fits into your overall marketing strategy.

AI jargon you need to know

Before we get into the details, it’s worth noting that many people use terms like “AI,” “machine learning,” “LLM,” and “generative AI” interchangeably. While they’re all related, they’re not the same.

Understanding the differences between each will not only make you the cool kid at the lunch table, but it will also help you make smarter decisions about how to leverage them in your marketing strategy.

The table below (and downloadable PDF) will keep you sharp and ensure you’re talking about LLMs correctly and authoritatively.

Artificial intelligence (AI)

Let’s start with the umbrella term. Artificial intelligence refers to the broad concept of machines being able to perform tasks that typically require human intelligence. This includes skills such as problem-solving, learning, speech recognition, and decision-making.

It has become a catch-all buzzword that gets thrown around the search industry and in headlines, but in practice, most of what marketers interact with falls under more specific subsets that we’ll get into in a second.

Example of AI:

Tools like Siri or Google Assistant use AI to interpret voice commands and respond contextually.

Machine learning (ML)

Machine learning is a subset of AI—it’s a process that feeds machines data and lets them learn from it. Instead of being explicitly programmed with instructions, machines learn using algorithms that find patterns and make predictions based on that data. (Sorry, it still doesn’t predict the future, yet!)

In marketing, this powers many things, such as ad targeting, customer segmentation, and predictive analytics.

Example of machine learning:

Netflix uses machine learning to suggest shows based on your watch history.

Natural language processing (NLP)

Natural language processing bridges the gap between raw text and machine-readable intent. 

NLP allows machines to understand, interpret, and generate human language. It’s the reason tools like ChatGPT can have a conversation, and why Google can know that “cheap running shoes” means the same thing as “affordable sneakers.”

Example of natural language processing:

Google Translate uses NLP to understand and convert languages in real time.

Generative AI

When people say “AI,” they usually mean generative AI, a branch of artificial intelligence that creates new content rather than just analyzing existing data.

At its core, it’s a type of computer model (built from algorithms and code) trained on massive datasets to learn patterns in language, visuals, or sound. Then it uses those patterns to generate something new, like text, images, code, or even video. (Google and Reddit’s partnership makes a ton of sense, now, right?)

However, it’s important to remember that it doesn’t “think” like a human. Instead, it predicts the most likely next word, pixel, or line of code based on probabilities, one step at a time.

That’s also why it occasionally goes off the rails.

These misfires, called hallucinations, are when the AI confidently makes something up. And they’ve become legendary. 

Among the most infamous examples: A chatbot recommending you eat at least one small rock per day, or that you use glue to keep the cheese on your pizza.

Examples of generative AI:

• ChatGPT writes articles and emails
• Midjourney and DALL·E create images
• Claude writes code for your WordPress plugin (I used Claude to write/build the entire SEO interview simulator on SEOjobs.com this way!)
• And lets not forget Sora, which I would never utilize to portray that I own a Ferrari 458 let alone do epic donuts. 

Large language models (LLMs)

Large language models are a specialized type of generative AI trained on massive datasets (think books, websites, code, and, yes, even Reddit) to understand and generate human-like responses. The bigger and better-trained the model, the more useful the output.

Think of LLMs as the engine behind your favorite chatbot, the part that responds to what you type.

When I first started using ChatGPT, one of the first things I did was upload years of my own writing: blog posts, newsletters, and articles on industry publications (like here on Search Engine Land). That context helped the model learn and respond in my voice, matching my tone and phrasing more naturally.

I still do the writing. The model just helps refine it by offering suggestions, rewording ideas, or polishing my draft based on the examples I’ve given it.

In short: LLMs don’t act on their own, they react to your input. That’s what separates them from AI agents, which can actually take action on your behalf.

Examples of large language models:

• GPT-4 (OpenAI, used for ChatGPT)
• Claude (Anthropic)
• LLaMA (Meta) 

AI agents

While LLMs simply respond to prompts, AI agents execute complex tasks—they navigate websites, fill out forms, call APIs, and complete multi-step tasks without hand-holding. They’re still powered by LLMs under the hood, but now they’ve got goals, tools, and autonomy. 

In other words, AI agents aren’t just talking, they’re working. 

And yes, these are what everyone’s afraid will steal their job.

Examples of AI agents:

• ChatGPT (Web searching and analyzing code)
• Google Gemini, Google Workspace (Gmail reply suggestions and email thread recaps)
• Copilot, Microsoft 365 (Microsoft’s version of Gemini+Workspace)

How AI impacts marketing today

Now that we’re all speaking the same AI language, let’s talk about how these tools are collectively disrupting marketing and the way businesses operate.

People have been claiming SEO is dying for over a decade, but this time, the anxiety is a little more concrete. And saying SEO is merely “changing” is an understatement. 

But we’re not facing death, we’re in the middle of a massive, industry-wide pivot, and AI is undeniably at the center of it all.

Let’s look at a few issues. 

Organic traffic is getting cannibalized

AI Overviews are Google’s automated summaries that appear at the top of search results, often pulling from multiple sources. 

Think of them as Featured Snippets on steroids—except they don’t just quote one site and link back. They blend multiple sources, rewrite them in Google’s own voice, and often bury attribution below the fold.

That means for broad informational queries, the first thing people see is Google’s answer, not your blue link. This leads to a much lower click-through-rate (CTR), which in turn reduces clicks to your website. 

Before AI Overviews, queries used to be a great way to introduce your brand to people early in their research, but now the answer and trust goes directly to Google. Sigh.

Claim: AI Overviews only show up for fluff queries, so my traffic is safe.

Reality: Google is already testing AI Overviews for YMYL, product, and B2B queries. It’s expanding, not shrinking.

Actionable next steps:

• Stop chasing every click. Focus on being the trusted source people remember when they’re ready to act. 
• Measure success by visibility, influence, and conversions, not just raw traffic.
• Double down on topical authority so you’re the brand cited in AI answers.

Content creation is exploding (and so is the noise)

Generative AI has removed one of the biggest bottlenecks in content marketing: time. What used to take a team of writers a month can be done by one marketer in a week. 

That’s not inherently bad, but when everyone can flood the internet with good-enough content, the signal-to-noise ratio tanks.

Claim: More content = more traffic.

Reality: This was shaky even pre-AI. Now, search algorithms are actively throttling quickly produced, low-value content. 

Google’s Helpful Content update, Bing’s SpamBrain improvements, and even LinkedIn’s recent feed tweaks are all aimed at burying generic, low-quality content.

Actionable next steps:

• Focus on authority-driven content: case studies, data analyses, and proprietary insights.
• Publish less, promote more. Distribution is more important now than ever before.
• Use AI for research, outlining, and refreshing content, rather than just producing more. 

Search results are becoming deeply personalized

Traditional SEO has dealt with personalization for years (e.g., local search, logged-in history), but LLM-powered platforms like ChatGPT, Perplexity, and Gemini are taking it to a whole new level.

The same question can return completely different answers depending on the user, their history, their prompts, or even how the model is trained. That means the idea of a “universal ranking” for all users is disappearing fast, and so is our ability to reverse-engineer it.

Example of personalization:

Here’s how an LLM might respond to two different users querying, “What is the outlook for Tesla?”

• Financial analyst: The LLM may provide a detailed stock performance summary and recent SEC filings, drawing from the user’s history of researching finance.
• New driver: The LLM may focus on new car models, battery life, and charging infrastructure, inferring an interest in buying a vehicle.

Claim: We’ll just optimize for the top answer in ChatGPT like we do for position #1 in Google.

Reality: The universal ranking is dead. The assumption that there’s one top answer to optimize for is flawed because personalization destroys the concept of a “top” answer. 

In classic SEO, Position #1 was a fixed, measurable target based on a universal algorithm (even if local user factors slightly adjusted the result). Optimization strategy was entirely built on this universal ranking model.

Actionable next steps:

• Track visibility, not just rankings—how often you’re mentioned, cited, or linked across different LLMs in addition to search engines.
• Build recognizable, credible brand entities that AI models can confidently cite.
• Invest in content variety (articles, podcasts, videos, datasets) to give models multiple ways to access your information and understand your topics deeply. This includes using structured data and creating definitive, citable answers that are easily ingested by LLMs.

Attribution is breaking

When Google, Bing, or Perplexity answer a user’s question directly—possibly using product comparisons, summaries, or “best of” lists—those users are likely to never visit your site, thanks to AI Overviews. Even if they do, their journey may start in an AI platform, jump to another search, and only then hit your site. 

This path shatters the neat channel → click → conversion path marketers have traditionally relied on, making companies rethink how to track attribution.

Claim: We’ll measure traffic and users from the LLMs directly in our analytics.

Reality: This claim is making a huge assumption that users are clicking to your site—they aren’t!

Actionable next steps:

• Shift from last-click attribution to assisted conversion thinking—last-click attribution gives 100% of the credit to the final channel before a conversion. Start looking at assisted conversions and multi-touch models so you can see which channels regularly show up earlier in the journey.
• Measure how people interact with your brand as a whole—instead of obsessing over a channel’s ROI, start tracking indicators of overall brand demand. A few concrete signals:
  • Direct traffic trends
  • Branded search volume
  • “How did you hear about us?” fields in lead/purchase forms 
• Budget for off-site influence that is hard to perfectly track, for example:
  • Podcast sponsorships
  • PR initiatives
  • Thought leadership that gets covered in news

Clients and bosses expect magic

Thanks to hype, stakeholders expect AI to do everything faster, cheaper, and better, often without understanding the risks, the learning curve, or the human oversight required.

Claim: We can replace our SEO/content team with AI tools and get the same results.

Reality: AI can accelerate tasks, but it’s not a replacement for strategy, judgement or truly understanding customers needs.

Actionable next steps:

• Set expectations early. AI can make some things faster and cheaper, but it’s not a push-button solution.
• Show stakeholders the hidden work: prompt refinement, editing, fact-checking, and compliance. (And that it’s not magic).
• Use AI’s wins to free up human bandwidth for the high-impact strategic work. The landscape is shifting fast and marketers will have to adapt to thrive. 

Search is evolving

I’m not here to argue about Generative Engine Optimization (GEO) vs. Answer Engine Optimization (AEO) or dive into the alphabet soup of SEO acronyms. What matters is this: Search today is not what it was yesterday.

Organic traffic isn’t just about ranking in Google anymore. The definition of “search” has expanded, and your strategy needs to keep up. If you’re not thinking beyond the search bar (YouTube, Reddit, newsletters, communities), you’re already falling behind.

The good news? We’re here to help. We’re publishing insights from industry veterans focusing on what to try, what to avoid, and how to actually talk about AI and search in your org without getting lost in the hype.