Insights
Gary Illyes Reveals His Stance on AI Content
I'm not sure Google knows how much AI-generated content they link to in Search.
So, when Gary Illyes (Google Analyst) was interviewed at the recent Google Search Central Bangkok event, I was intrigued by his take on AI content.
Illyes said:
…model training definitely needs to figure out how to exclude content that was generated by AI otherwise you end up in a training loop which is really not great for training. I’m not sure how much of a problem this is right now…
Illyes emphasized that "human curated" content (not "human created") is what matters.
“I don’t think that we are going to change our guidelines anytime soon about whether you need to review it or not. So basically when we say that it’s human I think the word human created is wrong. Basically it should be human curated. So basically someone had some editorial oversight over the content and validated that it’s actually correct and accurate.”
See the contradiction here?
Google's training models need to avoid AI-generated content to prevent "training loops" – but high-quality, human-curated AI content is apparently fine for training.
Someone needs to review and validate AI output, which is what every smart content marketer does.
But what exactly constitutes sufficient curation?
Illyes doesn't define this threshold, yet it's supposedly the difference between acceptable vs problematic content.
He also revealed several personal biases, admitting he "doesn't like generative AI" and worries about hallucinations when AI tackles unfamiliar topics.
When a key Google figure has such strong personal opinions, what does this say about algorithmic decisions down the line?
Here's my take:
- Keep focusing on quality over everything else (what's new?).
- Use AI as a drafting tool, not a replacement for human expertise.
- And definitely don't pump out unreviewed AI content—which you're not doing, are you?
Lastly, I'm keeping an eye on the ITF (Internet Engineering Task Force) AI preferences working group he mentioned, which is "not a Google thing."
This group is focused on developing a standard that would allow publishers to granularly control what their content can be used for by AI.
This could impact both content strategies and discovery in Search.
Bottom Line: We're all still figuring this out together.
Quoteworthy
Ignoring the 'GEO' acronym, here's what's important about the quote below:
- Brand marketing teaches you how to create memorable, search-worthy brands.
- PR gets your clients quoted and featured, building the authority that makes people search for them by name.
- Copywriting helps you craft content that ranks, persuades, and converts the traffic that actually reaches your site.
SEOs who can build brands people actively seek out will succeed because they understand the importance of all those signals.
Brand marketing, PR, and copywriting are among the most critical skills for SEOs to understand moving forward.
Good GEO will be about managing, building, and increasing the value of a business’s online reputation and presence.
But equally, that also means that SEOs can start to enter markets they haven’t worked in before.
Adding PR and copywriting to an SEO agency’s range of services is a natural step that many are doing.
—Andrew Holland in an article for Search Engine Land
25.8% of Web Pages Are “Pure Human.”
I'm skeptical about the accuracy of AI content detectors.
Still, this data from Ahrefs is worth a second look:
Using their inhouse AI content detector, Ahrefs analyzed 900,000 new web pages (one per domain) in April 2025.
- 74.2% of them contained AI-generated content.
- 2.5% of pages were categorized as “pure AI.”
- 25.8% were categorized as “pure human.”
- 71.7% were categorized as a mix of the two.
What does this tell us about the current state of web content creation?
- AI adoption is now mainstream in content creation.
With nearly three-quarters of new web pages containing some AI-generated content, we've clearly moved past the early adoption phase.
- Hybrid approaches dominate over pure AI.
Most content creators are using AI as an enhancement tool rather than a complete replacement for human writing. This suggests a collaborative approach to AI integration.
- Quality considerations may be driving the hybrid approach.
The low percentage of pure AI content might reflect creators recognizing that fully automated content often lacks the subtlety, accuracy, or brand voice that human oversight provides.
- Pure human content is becoming a minority.
The sad fact is that only about one in four new pages contains exclusively human-written content.
This represents a dramatic shift in how web content is produced and could have implications for content authenticity, SEO strategies, and reader expectations.
What does Google say about AI-generated content?
Google's documentation tells us that it's ok to use AI to generate content:
AI has the ability to power new levels of expression and creativity, and to serve as a critical tool to help people create great content for the web.
And,
If you see AI as an essential way to help you produce content that is helpful and original, it might be useful to consider. If you see AI as an inexpensive, easy way to game search engine rankings, then no.
The low percentage of “pure AI-generated” pages in the Ahref's data (2.5%) gives me hope, but that still represents 22,500 pages.
According to Tech Business News, around 7.5 million blog posts are published daily. If you take 2.5% of that number, you get 187,500 posts generated entirely with AI.
Every. Single. Day.
As more AI-generated content is produced, future content that draws on it will also contain AI-generated elements.
Logically, that 2.5% number will only increase.
And how does anyone source 'quality content' then?
Do you follow me on LinkedIn? I share regular tips and stories I don't have room for here. Come and join me.
