How AI is changing SEO as we know it
The one thing you need to know in AI today | AI Ready CMO
SEO expert Mark Williams-Cook sat down with The Neuron for a 77-minute conversation about AI and search. If you’re hoping for someone to explain why your website traffic is quietly bleeding out while AI tells confident lies about your brand, you’re in the right place.
The core problem isn’t that AI is replacing search—it’s that AI is breaking the feedback loop that made search work.
Google’s PageRank was built on a link graph: sites linked to quality, quality got visibility, users clicked through and reinforced the signal. But when ChatGPT or AI Overviews summarize your content without sending anyone to your site, that loop dies. Fewer visits mean fewer links, which means the link graph rots, which means Google has less signal to work with, which means search quality degrades for everyone.
To make things worse: LLMs hallucinate by design. Their reward functions incentivize plausible-sounding answers over “I don’t know,” so a 1% error rate at Google’s scale means millions of confidently wrong answers circulating back into the training data. The system eats its own tail.
Mark’s advice isn’t to panic or pivot to “Generative Engine Optimization” (which he dismisses as SEO with a new acronym). Instead, he doubles down on fundamentals.
He offers plenty of actionable advice in the interview, but three things stood out to us:
Stop doing keyword research; start scenario mapping. Instead of optimizing for “running shoes,” build personas—a vegan runner, a CrossFit enthusiast, someone training for their first marathon—and ask what theywould ask ChatGPT. Use the API to simulate those conversations. This isn’t about stuffing variations of a search term; it’s about understanding the conversation paths that lead to your product.
Get on Reddit and forums—but actually be helpful. Mark is blunt about this: don’t astroturf, don’t spam links, don’t “climb in through the window and talk loudly.” These spaces increasingly influence both AI responses and Google’s results, but they punish inauthenticity instantly. Provide real value, become the story that communities want to talk about, and participate like a human, not a marketing department.
Move your content into HTML. Most LLMs don’t render JavaScript, so if your primary content and navigation live in dynamic scripts, you’re invisible to the models training on your site. Talk to your tech team and get everything that matters into raw HTML.
And, if ever in doubt, Mark’s “Golden Rule” is a great guide: Never do anything for SEO that harms user experience. For instance, if optimizing for AI makes your writing worse for humans, you’re doing it wrong.
The uncomfortable truth Mark keeps circling back to: Google’s system goal isn’t “organize the world’s information.” It’s “make ungodly amounts of money.” Organic results exist to support the ad business, and publishers who threaten that model get algorithmically kneecapped. AI search is following the same playbook, just faster and less transparent.
So instead of chasing the next optimization trick, invest in the stuff that actually compounds: better product, clearer differentiation, stronger customer relationships. The models will eventually reflect that. The question is whether your site will still be around to benefit when they do.
3 AI Marketing Tools To Try Today
Tailwind
Schedule and optimize social media content across platforms—Tailwind automates posting, suggests best times, and helps you maintain consistent presence effortlessly.
Surfer SEO
Optimize content for search without guesswork—Surfer SEO analyzes top-ranking pages and gives you exact guidelines to outrank competitors.
Scalenut
Research, write, and optimize SEO content in one platform—Scalenut generates keyword clusters, outlines, and drafts that rank faster than manual methods.