Would You Use an AI Chatbot to Write Your Blog Post?
Many people already are.
Whoa, whoa. We need to back up a bit here.
Really good GPT-based text generators have now been generally available for 3 years now. And they’ve now been adopted by a large number of corporations, especially IT-centric ones.
I shouldn’t even say “adopted”. They’re being “pushed” because of worries that employees’ skills, and thus their organizations themselves, will fall behind in the skills race, or be less productive than the competition.
Perhaps however not every use of AI-based text generation is really necessary, or desirable. Let’s take an example dear to my heart: writing blog posts.
You’re Not (The Only One) Using AI-Generated Posts
People are writing their blog posts with AI. Are those posts any good? Would you sign your name to it if you did it?
The “helpful” information already out in the world isn’t as varied as I thought.
- Trying to sell something to ease the task of writing or be more prolific:
- Trying to sell something indirectly, like SEO courses:
- Actually trying to help you improve/clean up what AI wrote (though they may be trying to sell something, too)
- Or my personal favorite, an AI-generated video complete with AI text, synthesized voice, and slide layout to tell you how to use AI to write blog posts
A Quick Look at Blog Posts
An expert, human-written technical blog post definitely sounds like it was written by a person. You can probably even tell what they were thinking when they wrote it, like “Wow, he was trying to finish that as quickly as possible”, or “Yeah, she actually knows what she’s talking about, every detail you need to make IP tables in Linux is in there.”
The historical experience is evident, there’s a lot of detail, but not so much that half of it seems unnecessary.
There’s a definite target audience, and (since it’s technical) that audience is not the general public. And it’s definitely helpful NOW to that audience.
On the other hand, have you ever read a generic AI technical blog post? Here’s one I tried recently:
Me: What types of monitoring can NetEye do?
ChatGPT4: NetEye offers a variety of monitoring capabilities, particularly with its NetEye 4 update. Some key features include:
Performance Monitoring: It tracks the performance of IT systems, applications, and networks to ensure they operate efficiently.
Infrastructure Monitoring: NetEye can monitor servers, storage, and databases, providing insights into their health and performance.
…
These capabilities reflect NetEye's commitment to continuous improvement and adaptation to market needs. For more detailed information, you can check out their blog here.
It’s correct of course. But it also sounds like it was scraped from a web page written by Marketing.
Sometimes you have to look at more than one blog post before you notice what’s going on. They all read the same, and they have the same structure: paragraph, bullet points, paragraph, bullet points, paragraph.
Why it looks like this is the topic of an entire other blog post (which maybe AI can help me write), although here’s a pretty good, existing summary: https://medium.com/data-science-at-microsoft/how-large-language-models-work-91c362f5b78f
So What’s My Advice?
These systems will get better over time, of course. But for now, assume that the result was written by an intern who’s been at your company for two weeks.
That means you’ve got to use the text as a starting point. Don’t just take the output and copy and paste it into the blog.
I saw a post on the web that obviously hadn’t been proofread because it contained some text, then a section labeled “Conclusion”, followed by more sections, and ended with a second “Conclusion” section. Oops.
They are many types of technical blog posts, and they all have their specific goals and target audience. If it’s a promotional post, don’t include some example CLI commands. If it’s about Linux IP tables, you had better put some example config files in there. And they had better be correct.
Nothing will tank your blog’s reputation as fast as a technical article with examples, seemingly written by a human expert, but when the reader tries out those examples, they’re obviously broken or fake.
By all means use AI to get started. The best way to beat writer’s block is to have something that’s already 50% of the way there, that you can just critique and edit. It’s easier to add a paragraph than to write everything from scratch.
Sprinkle in some humor (inside jokes are great), add anecdotes that are highly related to the topic, modify the text to fit your personal writing style, insert formatting and images, and make appeals directly to your reader (you didn’t know you could do that, did you?)
Finally, you could be proud of your technical expertise. If you didn’t use AI, perhaps end your post with a line like this:
This blog post was 100% written by a human.
Charles Callaway
Author
