Flatpak

I ordered some furniture for my balcony some months ago, and today it arrived. It’s a small balcony, so I thought some small chairs would make sense…

The shipping company said they were going to be delivered in one 47kg package:

And I got heart palpitations, because I hadn’t ordered carry-stuff-up-all-the-stairs service, and how on earth could two small chairs and a small table be 47kg!?

It turned out they’d just strapped the three curiously, but efficiently packaged things onto a big wooden pallet, and the delivery guys fortunately took that thing with them.

Which left me with this IQ test — how do I get those black metal tabs onto the rounded leg section of the table? D’oh! That’s not how it works… rounded things go on the bottom, not the top. Hah! I’m a genius! It only took me *mumble* minutes to figure that out!

Et voilà! Now I can sit out on the balcony like a proper French man, smoking cigarettes and drinking coffee for breakfast while reading Libération!

(I just have to start smoking and drinking coffee first.)

I suddenly remembered that I had bought a battery powered reading lamp some years back… perfect for reading at midnight.

Finally a place I can read performatively in the privacy of my own home.

There’s something abhorrent about what AI produces

Here’s my little thought: Humans find AI output to be subtly disgusting and avoid subjecting themselves to it.

Because that seems to be the only reasonable explanation for stuff like this:

Again and again we see people having an LLM generate text and then somehow avoid reading before posting.

Sure, people are lazy, and laziness is the go-to explanation for everything, but:

We see renowned mainstream book publishers publish books containing phrases like:

“Of course – here you will find the text about artificial turf fields and ball pits in Norway, written in the same literary and discussion style as the rest of your project.”

This means that nobody — no bo dy — read this book before it was published. Not the “author”, not the editor and nobody at the publisher at all. And these are people who like reading stuff normally!

The same goes even more for pictures — I just selected a random example of somebody using an LLM to generate a picture, posting it, and then apologising afterwards when somebody actually looked at the picture and went “eeek!”

The only explanation that makes sense to me is that even the people who use LLMs to generate these things subconsciously find these artefacts to be so offputting that they avoid casting even an eye at the product. Just… “it’s there! get it out! out damned thing!”

What prompted this blog was this head-scratcher of a blog post: “AI-generated blog post images are not cool any more” They’ve never been cool! People have been calling them disgusting all this time!

It just goes to show how people really live in different worlds… You’re innocently scrolling through your feed and then bam! An AI drawing! Eww! Ick!

I’ve run thorough statistical methods on Hacker News and found that 72.3% of all posts end up with an argument about whether the article in question is written by an LLM. Which is understandable: There’s nothing as nausea inducing as reading a blog article and then having the slow realisation that you’re reading LLM slop. It’s led me to avoid reading blog posts that are about AI, because so many people in that sphere stop writing and just have the LLM generate the text.

But LLM output is abhorrent.

In conclusion: