Using AI to create a website- are companies like ours in danger of being replaced?

The short answer is “no”. Certainly not any time soon! Let me explain.

I recently received a promotional email from one of our website hosting companies. The specific one is not important since they’re all starting to offer this service, but it wasn’t the one that rhymes with No Baddy, it was another one. My first thought was “well, that’s not good!” since our company has been based on the act of making websites, specifically WordPress sites, for the past 11 years.  The idea of someone just pushing a button to have a site built for them was not a happy one.

But as they say, keep your friends close and your enemies closer. So I decided to use this new tool to build a website to see how it turned out. It did not go very well.

First off, I couldn’t find the tool! I have a hosting account with this company and it has many websites so where was the button for me? Turns out you don’t see the button until you get as far as installing an empty WordPress instance and you’re ready to start building the site. I only found that out after a chat with tech support too! Not a great start. Interestingly, once I got to the point where I was ready to build the site, there were three options- have AI build it (first button), do it yourself (middle) and “Hire an expert!” (third). So right off the bat this company is actually trying to eat into their own business.

So I click on the AI website builder button and it asks me to name the site. I decided to use the name of a current client to see how an AI version would compare with what I had already started to build. The second step was to provide a logo, if any. I did that, figuring that it would take the colors for the site from the logo. The last step was to describe the website in as much detail as possible. I did that, an entire paragraph about the site, the person that owns the business, etc. I also used some text from the clients’ current site which describes him and his business as well as anything.

And that was it! I clicked the button and off we went. After about 15 seconds, I was presented with three options and the ability to preview them all before choosing one, or starting the process over with three new choices. There was one that was “less bad” than the other two so for the sake of moving my experiment forward I clicked on the “this one” button or whatever it said and a little while later I was in the WordPress dashboard.

Here are my observations about the site it created-

• The color scheme was bizarre– despite having the logo in gray and blue, and most of the stock photos that were used being Chicago colors, the AI chose green and yellow as the main scheme. Very odd considering that’s the color scheme of our main rival in the NFL, the Green Bay Packers. Poking around on the site it was clear that there wasn’t much of a scheme at all. It was more like a box of random Legos had been used to build a house rather than a kit. To say there was a “color scheme” would be very generous.
• The stock photos were “CHICAGO” in every way– a Cubs hat, a skyline view, or just urban views. These had nothing to do with the business I had described to the AI at the beginning. It seemed all it read was that the business was in Chicago and away it went.
• It used the “Block Editor” for the theme– I am not a fan of the block editor for WordPress at all. I think it “Squarespace-ifies” WordPress which seems incongruous to me. WordPress is like Legos in the sense that there’s a piece of code for just about everything you want or need to do. For me, Squarespace is like Duplo- much more basic and restrictive.
• The AI chose the theme– I like to use certain themes over and over because I have gotten to know them pretty well. The AI just seemed to pick a theme that had a name similar to the vocation of the website. That’s not a good reason to use a theme.
• It was “custom” in the sense that it contained some of my text, but not in any other way– the AI decided to install plugins for selling items through the website despite never being told this was necessary. So it had things we didn’t need, but missed simple things like a contact form or phone number to contact the business.
• It would take a lot of work to shape the site into something acceptable– honestly it would not have saved much time over what we currently do which is install demo content and go from there.

In the end I was pretty relieved at what a terrible job it did “making a custom site”. The whole experience reminded me of an exercise from one of my college journalism classes. The instructor asked us to describe how to make a peanut butter and jelly sandwich to an alien that had just arrived on earth. Sounds pretty easy until you realize that the alien doesn’t understand the concepts of “eating”, “sandwiches”, what peanut butter or jelly is or why you like it in a sandwich, what “bread” is or how many versions of it there are in the world, where any of the ingredients come from or how a tool called a “knife” which can sometimes be used to injure someone is also used to make this “sandwich”. It’s a lot! There’s so much subtlety and actual complexity in something like making a sandwich. That is exponentially increased with something like a website, which is not one thing but many different things joined together.

There might be a day when this is a real useful tool, but at the moment humans have nothing to fear with certain creative tasks.