Scaled AI Content vs People-First Content: Is AI Generated Content Good for SEO?

Google's spam policies define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users." The first example Google gives is "Using generative AI tools or other similar tools to generate many pages without adding value for users."

Read that sentence closely. The problem it names is the combination of volume, ranking intent, and no added value. AI is listed as one way to produce that pattern, not as the offence. Google applies the policy to content "no matter how it's created," so a thousand thin pages written by hand would fit the same description.

What this means for you: the question to ask about any page is not "who wrote this?" but "what does this page give a reader that they could not get elsewhere?"

Google's guidance on generative AI content is direct: "Appropriate use of AI or automation is not against our guidelines." That puts AI drafting, outlining, translation and editing inside the rules, as long as the output serves people.

Google's helpful content guidance describes people-first content as content made for a specific audience, not for search engines. It also says E-E-A-T "isn't a specific ranking factor." There is no score to raise by adding an author bio or credentials block.

So the practical line is simple. Use AI wherever it helps you produce a better page faster. Don't use it to produce more pages than you can make worth reading. For a fuller look at the penalty question, see does Google penalize AI content.

The last row matters for AI search. Google states that a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode. Eligibility is not inclusion, but losing the index removes the option entirely.

OtterlyAI ran a public test in 2026. It launched two new domains about capybaras, each loaded with roughly 1,000 AI-generated posts in one click. One site, capybara-world.com, showed an author box presenting a fictional wildlife biologist. The other, capybaraverse.com, had no author.

Google removed both sites

• capybara-world.com lasted about 14 days in Google's index. Its daily impressions fell from 1,629 to 15 on April 9.
• capybaraverse.com lasted about 90 days. Its daily impressions fell from 859 to 85 on June 26, then to single digits.
• Otterly reports zero manual actions in Search Console on either domain. The removal was algorithmic and silent.

The site with the fake expert went first. If author boxes shielded scaled content, the order would have run the other way.

AI search split by index

Bing never deindexed either site. Of the 3,934 AI citations Otterly tracked for capybara-world.com, 2,718 came from Microsoft Copilot, which grounds on Bing. Google AI Overviews, AI Mode and Gemini almost never cited the sites. Our breakdown of the experiment's method and limits has the full numbers.

What this means: an engine can't cite what its index no longer holds. Copilot citations on a low-stakes topic did not replace losing Google.

Otterly published its data, which is more than most GEO advice does. It is still a small test, and the claims going around stretch it.

• One site per arm. The two sites differed in domain name and more, not just the author box. The author result is one data point.
• No human-written control. Without a site publishing 1,000 human-written posts overnight, the test can't separate a reaction to AI text from a reaction to sudden volume on a domain with no history.
• A low-stakes topic. Capybara questions face little competition. The result says little about commercial categories.
• Prompt-dependent citation counts. A tracker counts answers to the prompts it asks, and the prompt list isn't public.
• A vendor ran it. OtterlyAI sells AI visibility tracking, and so does 99Visibility. That doesn't make the data wrong. It is why replication matters.

Treat the test as a clear illustration of Google's written policy, not as a new rule.

The fix is a review step, not a ban on tools. Before you publish AI-assisted pages in bulk, check each one against a short set of questions:

1. Does this page answer a question a real customer asks?
2. Does it contain something specific: a price, a number, a process, a one-line definition you can stand behind?
3. Have the facts been checked by someone who knows the subject?
4. Would you publish it if search engines didn't exist?

A pre-publish review for scaled AI content turns this into a repeatable process, and a fact-check workflow for AI writing covers the accuracy step. Stating specific facts on a fetchable page is also the one on-page practice with any positive experimental signal for AI answers: it gives an engine a sentence it can quote. That is not a promise of citation.

Watch volume too. Publishing more does not by itself earn visibility, as covered in whether publishing more content helps AI visibility.

Use AI-assisted, people-first content when:

• You have real expertise or data and need help drafting, structuring or editing it.
• Each page targets a question you can answer better than what already ranks.
• You can review every page before it goes live.

Avoid scaled AI content when:

• The plan is to publish hundreds of pages to cover keyword variations.
• Nobody will check facts before publishing.
• The domain is new and the content has no source beyond the model.

Check after publishing, either way:

• Look at Google Search Console and Bing Webmaster Tools separately. One being healthy tells you nothing about the other.
• Measure each AI engine on its own. In the Otterly test, one engine cited the sites thousands of times while four barely did.

If you are weighing tools, our guide to AI SEO content generators and their risks covers what they do well and where they fail. Whatever you choose, nobody controls what the models say. You control whether each page is worth indexing.