On maintaining a technical blog in the age of conversational AI

I created this website five years ago because I thought it would be nice to have my own little corner on the Internet where I could write and publish my technical notes, as a way to contribute to the community.

I feel that I have largely achieved that goal. Until now, the number of visits to my website has steadily increased year after year. A significant portion of traffic to Chuniversiteit currently comes from search engines, presumably by people in search of a solution to their technical problems.

But for how much longer will this continue?

Conversational AI services like ChatGPT, Bard and Bing Chat are nothing short of revolutionary. ChatGPT, in particular, has become incredibly popular in a very short time – not just among tech enthusiasts, but also among the general public.

Large language models like ChatGPT have the remarkable ability to provide almost real-time, tailored responses to any question or prompt fired at them. There is little point in trying to “compete” with something like ChatGPT as a technical blogger: when it comes to answering the kind of technical questions that most people have, an AI like this is nearly impossible to beat.

It’s true that ChatGPT makes mistakes from time to time. However, I’m confident that the quality of responses will improve over time (side note: Many who are sceptical about conversational AI as technology seem to overlook the fact that this stuff is brand new. There’s plenty of room for improvement!). Even in its current somewhat buggy state ChatGPT manages to be genuinely helpful, as most people are perfectly capable of recognising and handling incorrect responses.

What about “If you can’t beat them, join them”? Not only could I ask ChatGPT for article ideas, I could even ask it to do research for me, and write paper summaries, programming tutorials and clickbaity columns. To save even more time, I could use generative AI (Midjourney or DALL·E) to generate illustrations. But this would completely defeat the point of having a personal blog…

So this is what I’ll be doing from now on:

• Conversational AIs are excellent at explaining concepts on request, but can’t do much on their own: users have to manually “pull” information from the chatbot. As a technical blogger, I can “push” information to readers, e.g. to create awareness about certain problems, solutions, or other topics. If readers want to know more, they can always ask ChatGPT or good ol’ Google for more information.

• ChatGPT is trained with a dataset that’s absolutely massive, but not complete. It won’t know anything about information that was published only recently or in obscure sources, like an issue tracker or someone’s personal blog. Technical blogs still have an edge here.

• A personal blog is “personal” largely because it’s about personal experiences. I do a lot of cool stuff that I want to share with the world, and I think there are many enough people who’d rather read about that than some generic response from a bot.

• I’m not a native English speaker, so I make weird little linguistic mistakes all the time. This means I am also bad at proofreading my own work. Fortunately, ChatGPT can spot the most egregious errors in my writing.

• I’ll still create illustrations manually using Microsoft PowerPoint. However, not every article needs a brand new illustration. And short articles shouldn’t even have an illustration at all.