Should You Publish AI-Generated Content

(Disclaimer: Profanity.)

Yes, both individual bloggers and content marketing teams of businesses can publish AI-generated content and get some decent ROI. 

But only sometimes. 

There are a handful of cases when this can work without any repercussions.

Articles that contain definitions and descriptions.

Informative articles, messages that fit into the ‘just fyi’ category, press releases, organizational statements, product update log entries, etc., are the kinds of content that can be made quickly and efficiently with LLMs.

Glossary-type articles too. 

In the rest of the chapter, let’s dive into the advantages and disadvantages of publishing AI-written content.

Advantages and disadvantages of publishing AI-written content

I need about 2 hours and 30 minutes to write a ~1200-word article for a general internet audience.

It takes about 30 minutes to write a ~1200-word article by using my end-to-end ChatGPT content writing workflow. The custom GPTs aid you in every phase of content production and the process allows you to personalize the output to get useful drafts.

If you have a decent enough domain authority within your niche, you will rank on the first page of the internet.

Keep in mind that if your intention is to spam, it will be treated like spam. Visitors will drop off soon and the audience’s trust will erode. Hence, only produce content with AI if it fits the description I shared in the introduction of this chapter.

The disadvantage of LLM-produced articles is that they fail to influence the reader. The core tenet of content marketing is perception engineering. You have to talk to the reader directly to engage with them and motivate them toward a certain decision.

I’ve wasted many man-hours on these AI chatbots trying to get them to mimic the empathetic and engaging tone of a human. The truth is — it isn’t possible. 

Persuasive writing requires you to tweak the balance between subjective emotionality and factual objectivity. The sentences in a typical human-written paragraph flow dynamically which is the result of a complex interplay of many things.

AI simply can’t do that. 

How Google treats AI-generated content

Most bloggers and content marketing teams don’t give a shit about whether their article has ‘persuasive capacity.’ They just want to see the organic traffic go up, so they can rake in the desired ad revenue or get higher budgets approved from their superiors.

And I totally get it.

The truth is, in the earlier days, I’ve published quite a handful of AI-generated content for one of my clients. They did perform well and the client was happy with the boost in organic traffic. 

However, the dwell time on these articles was lower compared to the articles I wrote for them.

In the broad scheme of things, it didn’t matter as they were a local service-based business and just wanted to bump up their ranking in Google’s listings. But it got me thinking — this would have hurt a digital-first or digital-only business.

A lower dwell time coupled with high bounce rates results makes your website look bad in Google’s eyes. It simply means people don’t value your content enough to stay on your pages for long enough.

If this happens enough, it will incur a manual penalty from the overlords in California and your organic traffic and domain authority will nosedive. And a manual penalty from Google will brand you forever and the best way forward is to start from scratch.

There were many smartasses who thought they would get away with this but ended up losing most of their organic traffic and keywords position. [1]

As I suggested in the introduction, just use AI for content that’s not supposed to persuade or engage with your audience in any way. Think of celebrities uploading PR announcements on their social media channels.

Conclusion: The purpose matters

Machine-written content won’t earn you any readers, let alone fans. LLMs are programmed to be helpful assistants that fetch context-specific information and suggest ways forward. The text produced by these models has little emotionality which is crucial for engaging with the reader.

At the same time, there are plenty of cases where AI can do good enough of a job such as explaining company policies or providing word explanations. Before delegating a task to AI, just ask “Do I need the reader to believe or do something?”

—

Next Chapter: ChatGPT o1-Preview: Great for Content Planning

Previous Chapter: Four Limitations of the AI Content Writing Workflow

Index (with Prologue): Content Writing With AI

References

1. Google destroys AI spammers

Should you publish AI-generated content