Blogging in the age of AI

Table of Contents:

• Introduction
• My tendency to tinker
• The value of communication
• The current state of blogging
• Influence of generative AI on blogging
• Conclusion

Introduction

I'm writing this post to belatedly introduce myself and answer two questions:

1. Why did I launch a blog in a hurry, then forgot to update it?
2. Why would I launch a blog at all in the age of AI?

The first question is easy to answer: I went back to school after years of working in software development and studying has been intense!

The second question - why launch a blog at all in the age of AI - will take more time to answer. To do so, I'll consider my tendency to tinker, the value of communication, the state of blogging, and the influence of generative AI.

As we'll see, launching a blog in the age of AI still seems like a decent choice for self-expression with limited downside, understanding that blogging keeps changing.

My tendency to tinker

I grew up in the early DIY days of the World Wide Web, when GeoCities was very much alive and kicking.

Possibly related: I feel there's something to say for personalized websites, pieced together by eager amateurs who are figuring things out as they go. This can-do spirit reminds me of being a kid when, frankly, life felt a lot more fun.

Also possibly related: I still enjoy tinkering and am fine with personal websites being curiously imperfect. It's nice being able to curate an experience from the ground up and filling a canvas with anything from text to graphics to the occasional interactive tool, even when not practical in the long run.

That said, while I have built many websites in my life for different purposes, I've never written a blog. That feels strange to say because I've spent a lot of time writing in some capacity or another. However, the writing has usually been to pitch a project or troubleshoot a problem rather than communicating for its own sake.

The value of communication

Returning to school has made me reconsider the value of communication, both in my life and career. Writing itself hasn't been the issue, but I've come to see room for improvement in organizing my thoughts and building a decent narrative along the way.

I also feel like there's a lot of value in trying to find my own voice in middle age. It could be through social media but I find their feeds too frantic and impermanent for my liking. It could be also poetry or short stories, but my brain is still too full of technical details for those to feel natural.

Writing a blog seems like a great way to organize my thoughts and find my voice. It also gives me an excuse to learn about topics like meditation, school, and blogging itself. Plus, it comes without the added stress of needing to provide APA-style citations!

As mentioned above, though, I've never been a blogger. Are blogs still a thing?

The current state of blogging

Platforms

Blogs have come a long way since the early days of GeoCities. Many of those early weblogs moved to managed services like Blogger, LiveJournal, Movable Type, and the juggernaut known as WordPress which offers both the software to self-host a blog and a managed service. It has become so popular for personal and commercial use, in fact, that WordPress is used by 43.4% of all the websites.

In addition to WordPress, more recent services like Medium and Substack launched with highly polished user experiences and subscription models to help writers get paid by their followers. I decided to stick with these three major players and find data about their website traffic.

Website traffic

We've already established that blogs running on WordPress get a lot of traffic and that much of this traffic will be for commercial blogs backed by marketing teams. Since my aspirations are a lot more humble, I focused on blogs without a custom domain name.

I wanted to determine how website traffic to these blogs on Wordpress, Medium, and Substack changed over the last six months. Being very new to this, I wanted free website traffic tools and found ahrefs (good) and SE Ranking (less good for due to usage limits).

It seems that free data is hard to find and paid data is more expensive than I'd have guessed. All the same, here's what I found:

These numbers are a small fraction of the total website traffic to blogs, but I assume the overall trend still applies. If that's true then traffic is still increasing which is good enough for me and brings us to the community aspect.

Community

What's a blog without its readers and their feedback?

You wouldn't know it looking at this particular blog, but I know the real value of a blog is communicating with someone itself, and hopefully enough someones to get some conversation going.

Comments on blogs are an easy way to do this and still easy to find, but it feels like communities have largely migrated to social media platforms. It makes sense that people want to go where the party is even, if I find some of these parties a bit much..

All told, community still seems to be alive and well although I'm not sure how its migration to social media has changed blogging itself. However, I suspect the influence has been less pronounced than the influence generative AI will have.

Influence of generative AI on blogging

Market trends

According to a recent report, corporate AI investment reached $252.3 billion in 2024 which almost defies reason. While it's hard to see how that will slow down, and history is littered with market cycles that boom and bust, AI has proven useful in areas like automation.

Automation

My experience with blogging is on par with my experience using generative AI. In other words, sorely lacking.

Still, I appreciate the benefits of automation from a software perspective and generative AI using ChatGPT or similar promises plenty: writing outlines, drafts, and even articles themselves. It's hard to repeat the same task 100 times and not feel that some of it could be standardized, saving time for the unique parts.

Like all automation, generative AI promises to replace human labour with something faster (if not always initially better) and one of its first impacts is changing how existing automation works, specifically programmatic SEO.

Programmatic SEO

In the absence of webrings and in addition to direct referrals, websites have long relied on Search Engine Optimization (SEO) for drive website traffic based on meaningful descriptions, keywords, and valuable in-page content itself.

Optimizing these values across 10s or 100s of pages takes a lot of work and, at a certain scale, most websites (including blogs) will start filling variables into templates like:

%BLOG_TITLE%

Published by %BLOG_AUTHOR% on %BLOG_DATE%

Programmatic SEO refers to extending this approach across a huge number of largely identical pages, such as a travel website listing hotels or an e-commerce website listing products. In these cases, a consistent look across pages is a feature. My understanding is that this approach has been less effective for writing longer-form content (such as blogs) which doesn't lend itself well to templates.

However, generative AI has made it ridiculously easy to submit a prompt with a few keywords and a request to return a descriptive paragraph or five. This is where we get into concerning territory on two fronts.

First, the average blog relies on being discovered through Google or similar. As search results increasingly return AI generated content, there's a real risk of legitimate blogs (i.e. with human authors) being drowned out. I definitely ran into this searching for a mantra generator prior to creating my own.

Second, the large language models (LLMs) that drive ChatGPT and Gemini largely rely on data that have been scraped from the web (via search results) for their responses. These responses are far from perfect and there's obvious risk in feeding them back into LLMs through website content generated en masse. This concern is called model collapse and was flagged by researchers in a 2024 paper called "AI models collapse when trained on recursively generated data.".

While this line of reasoning might point towards the conspiratorial dead Internet theory, that's far too cynical for my liking. Still, some of its broader concerns like automatically generated content seem to be very relevant so it's worth considering how a web that's optimized for generative AI might look.

Generative Engine Optimization

One such trend is known as Generative Engine Optimization (GEO) and its goal is citations by AI's summarized responses rather than traditional clickthroughs. It's not hard to imagine the benefits to AI companies for websites to surface LLM-optimized content. What's less clear to me is how content creators will benefit, particularly when AI companies aren't generally paying for source data in the first place.

That said, it's great to see AI responses increasingly link to source material. This reinforces the reality that LLMs are summarizing and synthesizing what has already been painstakingly created by people and made public, which includes blogs.

Conclusion

I can only imagine how the web will continue to evolve but it feels like blogs will continue to have value. The medium provides a great opportunity to tinker, communicate and build community so long as traffic holds, so hopefully bloggers around the world will benefit from AI as much as they're impacted.

That said, I could easily be wrong and the internet might already be dead. If that's the case, I'll enjoy the art and craft of extending the web one page at a time, and resign myself to shouting into the void.

If you got something out of this or have feedback, please let me know (and thanks).