Is AI-Generated SEO Content Actually Good? Here’s How to Do It Right

The reality is, the influencers are lying to you

I'm pretty sure you've come across videos from "AI-Expert " influencers making incredulous claims --

"Create content with AI in 1 click!"

"How I Published 1000 AI SEO Articles in 2 Mins (AI Tool)! 🚀"

"How I Make 1,000+ Blogs In One Click (+SEO Optimized)"

If you are new to SEO, it could make you feel like the gates of heaven has opened (cue choir). SEO could be so easy! You thought.

But if you've been active in this space in the past year, then it shouldn't be news to you that Google has been rolling out huge penalties for junk-AI content since 2024. Plus, if you actually read any of the stuff that is generated through 1-click magic, it should be clear that no one will want to read that stuff.

Fundamentally, SEO is about creating value for your readers. If you can outsource it completely to AI, then chances are, your readers can also completely bypass you.

The good news is, AI still does allow us to take some shortcuts in our SEO efforts -- if we liken our previous manual approach to walking, then using AI should be like driving. How should one use AI for SEO? Let's dive in.

When it's OK to completely rely on AI for your content

Apart from generating long-form editorial content, part of SEO is about making sure each page of your website has fitting content. These include things like:

• Page descriptions

• Product descriptions

• FAQ sections

• Paraphrasing chunks of text

The common characteristics of these content, are that:

• You are able to articulate the key points, but it just takes a lot of time to flesh out bullet points into proper sentences.

• ...

In these cases, setting up an AI automation that transforms your limited input into coherent sentences that are both reader and crawler friendly can be a huge time saver.

Case in point: how I used AI to generate hundreds of page descriptions for my startup

When I was running my startup, the platform we built had hundreds of "category pages" -- these are pages that are like sub-pages to a directory. Each page holds a collection of data items under a shared theme. To write a page description for each page manually would drive me nuts, and will take up way too much time. So I set up a relatively simple automation with Make and Google Sheet, and generated these descriptions in a day.

Work in batches, and do manual quality-control

Even though the actual generation took just a day, setting up the automation took longer. In short, try to work in steps and small batches to assess the quality of AI's output first, before bulk-generating everything at one go.

When it's NOT ok to completely rely on AI for your content

I have three golden rules:

1. Everything where your unique perspective is what makes the content valuable, you shouldn't just leave to AI.

2. Everything that requires specialist knowledge -- but it deep technological-, industry- or company-specific knowledge, you DEFINITELY shouldn't just leave to AI.

3. Everything news related. It's simply irreponsible to generate news out of thin air.

When I first started out, I was experimenting with using AI to generate whole blog posts for my startup. (After all, that's the "holy grail" that all these influencers promised!)

Unfortunately for me, my start up is in the business of deep tech. Even for simple articles, 80% of the references in those articles were conjured from thin air (i.e. AI hallucination).

What's worse, was that ChatGPT was so good as disguising them as legitimate URLs from McKinsey.com and the likes, there was no way to see which URL was actually real, other than clicking through them one by one.

I ended up hiring proper tech writers that have PHDs. This approach ended up being much better for my SEO...more on that in a future post.

That was 9 months ago, and you might think that AI models have gotten so good by now that this would stop being an issue. Well in 2025 alone, there have been 73 cases where lawyers in the US including "AI hallucinated quotes, fake cases, or cited other apparent legal authorities that didn't exist." (see the database from Damien Charlotin here.) So don't be fooled.

Use AI for inspiration and quick research to augment your original thought

Here's what I actually use AI for, when I do content writing:

1. Generate ideas on what to write about.

2. Generate ideas for blog post titles.

3. Create a draft outline, just so that I can quickly get a sense of what aspects I should cover in the article.

4. Summarize an article for meta description.

5. Refine my writing, usually to look for a better expression or to make my writing more concise.

At no point does anything go out without at least a read-through and some form or manual editing. All reference links are checked manually as well.

These might seem mundane and even tedious, but let me tell you why this is important:

1. AI-generated articles all sound the same.

2. If you keep at it for a while, it becomes pretty easy to spot AI-generated articles (or even sentences) just from the choice of vocabulary, article structure, and use of emojis alone. Do you really want to sound like everyone else?

3. You are doing your part to not add to the AI-slop.

4. Thank goodness for you. There's hope for humankind.

5. You are doing your brand a huge favour.

6. Most companies do SEO because it's a way to pull potential customers onto the website. Imagine going through all that effort to lure customers in, only to irk them out. That's a colossal waste of your limited resources.

So conclusion: AI-generated SEO content can be good for limited use-cases

For "accessory text", go ahead and automate them away (with human supervision, of course). But for the longer, more thoughtful content that makes your brand shine, the human touch is really the only way to go.