This post may contain traces of AI

Armin Ronacher has an AI transparency statement on his blog.¹ He uses AI mostly as an editor or sparring partner, while he owns the “substance and original thinking”. I find this pretty cool and want to be transparent too.

It reminds me of reading the ingredient list on food items. I’m quite used to that, since I have celiac disease. That means I’m on a lifelong gluten-free diet and constantly scan ingredient lists for gluten. Some items don’t list gluten as an ingredient, but may contain traces of it. That’s often the case when something is naturally gluten-free, but might have been contaminated during production. Lentils, for example, might contain wheat grains from harvesting or processing.

If I were on an AI-free diet, I’d appreciate having that information available. I also wouldn’t want to read a fully AI-generated blog post. I want to read about other humans’ thoughts and feelings. But what if AI was only used to some extent, like a sparring partner? Then it’s not AI-generated, but it may contain traces of AI.

Eating something that may contain traces of gluten is a gamble. There are celiacs (like me) doing it and others avoiding it. I like to think that producers try their best to avoid contamination and rarely fail. That is what can happen as well when you use AI as a sparring partner. In many cases it might act as an amplifier or filter, amplifying or purifying the original human signal. But it might sometimes, unnoticed by the writer, distort that precious signal.

While I don’t mind traces of AI, I avoid fully or heavily AI-generated content. Others might want to avoid traces too, and some might even prefer fully AI-generated work. I don’t judge any of these perspectives, I want to respect them. That’s why I created an AI transparency statement, so you can decide for yourself whether my approach fits your preferences.