AI-generated content is already everywhere on the web. Ahrefs analysed 900,000 newly published pages and found that 74% contained some AI-generated content, while only 26% were classified as purely human-written.
We see the same thing in our own work. AI is already part of how we think through ideas, research topics, structure content, and get to a first draft faster.
Naturally, that leads to one question we hear from clients quite often: “The content is showing a high AI-generated score. Should we really publish it?” And we get it. There was a time when using AI for content felt like something you were supposed to hide, or that it made the work less credible.
But does a page become less useful because AI helped create it, or more useful because a human typed it? The better question is whether it adds something: original research, primary-source evidence, better analysis, or a more current answer. AI has made producing content much easier, but it has also made producing more of the same much easier.
Google’s scaled content abuse guidance makes the distinction clear: what matters is whether content adds value, regardless of how it was created. The more useful question, then, is whether the content actually helps the reader or adds another commodity version of information that already exists.
Key takeaways
- AI can help you move faster, but it only matters if the content still has something useful to say.
- Generic content is the bigger problem. It can come from AI or from a human rewriting.
- Strong AI-assisted content usually has better inputs: primary sources, current information, expert context, original data, or clearer analysis.
- A high AI-generated score does not tell you whether a page is worth publishing. Accuracy, usefulness, and what the page actually adds are much better signals.
What is commodity content?
Commodity content answers the topic without adding much that is new, specific, or useful. It may be accurate, but it gives the same information that can be found somewhere else.
In practice, it often looks like:
- Repeating the same definitions already ranking in search
- Summarising what competing pages have already said
- Recycling the same recommendations and examples
- Relying almost entirely on secondary sources
- Making broad claims without evidence
- Stretching a simple answer to fill a word count
- Answering the keyword without helping the reader make a better decision
AI gets associated with generic content because, with a broad prompt and weak inputs, it can produce a polished version of what already exists in seconds.
But that does not make commodity content an AI problem. A human can do the same by rewriting top-ranking pages. AI-assisted content can also be useful when it is built on primary sources, current data, expert input, client context, real examples, and proper verification.
How AI has changed content creation
AI has made a lot of the work around content faster. Research can be organised more quickly, sources can be compared in minutes, frameworks can be built sooner, and first drafts no longer start from a blank page.
A 2025 study on AI-assisted professional writing found the same pattern. Graduate students completing a research and writing task reduced their median completion time from 150 minutes to 65 minutes with AI support, a 56.7% reduction.
That extra time gives us more room to focus on the parts that are harder for AI to replicate:
- Expert and first-hand insight
- First-party data and original research
- Primary-source evidence
- Editorial judgment on what is worth answering
The content challenge is moving for us too. The part that still needs real thought is why a page should exist in the first place. Producing another page around a keyword is easy but what matters more is whether the topic, evidence, and answer will genuinely help the person reading it.
Can we produce something worth choosing, saving, citing, or acting on?
That is a much better use of the extra capacity AI gives us.
What actually makes content non-commodity?
Useful content usually creates value in one of three ways: it adds something new, does more with the information already available, or makes the answer more relevant to the person searching.
Add original value
First-party data, original research, expert input, customer findings, and first-hand experience can add something the existing results do not already have.
It could be a client sharing what they have seen in practice, a survey result, an internal benchmark, or a pattern that has emerged across customer conversations. A strong expert explanation can also add value when it gives the reader context they would struggle to find anywhere else.
Google’s people-first content guidance asks similar questions, including whether content provides original information, research, or analysis and whether it adds substantial value compared with other pages in search results.
Strengthen the evidence
Original value can also come from doing more with information that already exists.
- Going back to primary sources, checking whether commonly repeated claims are still correct, comparing conflicting information, and explaining what the evidence actually means can make a page far more useful.
- This is especially important for topics where details change often. Pricing, regulations, product capabilities, tax rules, and policy updates all benefit from current sourcing and careful verification.
- Good synthesis matters too. Bringing several reliable sources together in one place can save the reader from opening ten tabs and working out the answer themselves.
Make the answer more specific
Specificity makes an answer more useful when it reflects the reader’s actual situation: their country, industry, role, use case, or decision. That can make the content feel genuinely different too. A recommendation shaped for a first-time buyer, for example, should look different from one written for an experienced procurement team.
The value comes from making the answer more relevant, more distinct, and easier to apply.
The contribution can take different forms: stronger evidence, clearer interpretation, more relevant context, or better synthesis.
What ReSO’s client data shows about AI-assisted content
We’ve been using AI across content creation for multiple clients, so we’ve had a chance to see what actually happens when you scale it beyond just a few pages
For one client, an EOR platform, we created around 80 blogs over a four-month period. We focused on useful, specific content backed by primary legal and government sources. We also brought in current information and client expertise, which helped strengthen the E-E-A-T signals.
As the content started to gain traction, blogs became a much more important part of the site’s total organic performance. By the end of the period:
- Blogs accounted for 59% of the total clicks
- Blog clicks were up 26% month on month
- The entire net monthly gain in site clicks came from the blog library
- Newly published pages were already earning clicks in their first month
We are also seeing a similar pattern with a client in the healthcare space. Over a recent 28-day period, blog clicks increased by more than 7x and impressions by nearly 9x. AI was part of the production process, giving us more capacity to focus on stronger sourcing, better context, and content with a clearer reason to exist.
How to decide if AI-generated content is worth publishing
When a client asks whether a page with a high AI-generated score is ready to publish, we look at the quality of the finished page. A useful publishing check, which we also follow:
| Check | What to look for |
| Contribution | Does the page add useful information that was not already readily available? |
| Accuracy | Are the important claims sourced, verified, and current? |
| Depth | Does the page include evidence, expertise, analysis, or useful specificity? |
| Usefulness | Does it help someone understand the topic, make a decision, or take action? |
| Clarity | Is the answer easy to follow and specific to the subject? |
| Writing quality | Does the language feel natural, direct, and shaped by the topic? |
Writing quality matters here too. AI-assisted drafts can sometimes fall into repetitive structures or templates, vague transitions, or familiar phrases that appear across many pages. We edit those out because the final content should feel specific to the subject and useful to the person reading it.
The strongest pages usually have a clear reason to be on the web. They make something easier to understand, verify, or act on, and that is a far more useful publishing signal than an AI-generated score.
If you are already using AI to produce content, the next step is making sure that content is helping your brand show up where buyers are actually searching and asking questions.
ReSO helps you see where your brand is being mentioned, cited, or missed across LLMs, and where stronger content could have the most impact.
Book a call or get your free AI Search Audit to see where your visibility gaps are.
Frequently Asked Questions
Does Google penalize AI-generated content?
No, Google focuses on whether content is helpful, original, accurate, and valuable. Its guidance specifically targets scaled, low-value content regardless of how it was created.
What is commodity content?
Commodity content repeats information that is already widely available without adding meaningful evidence, expertise, analysis, specificity, or a clearer answer.
Can AI-assisted content rank in search?
Yes, AI-assisted content can perform well when it is accurate, well-sourced, useful, current, and built around a genuine information need.
What should humans add to AI-generated drafts?
Human input is most valuable for verification, expert context, first-hand insight, primary-source research, editorial judgment, and removing generic or repetitive language.
Are AI content detector scores useful?
They can indicate writing patterns, but they are not a quality metric. A high AI score does not tell you whether the content is accurate, useful, original, or worth publishing.
