The tech industry was once again rock solid in its conviction and this time I didn’t have the slightest doubt either. OpenAI’s models were so good that calling it a gold rush was an understatement for what I was seeing. Anyone could write entire texts with simple, concise prompts that sounded good enough to put down the pen for good.

I got caught up in it too. But there was a flip side. Because the same people who praised this wonder of technology as a cure-all were writing their own stuff by hand. I saw it again and again. I also kept writing manually and thought I was just slower than others. But many, really many, kept writing their content by hand.

I observed this for a long time. Consultants recommend AI content to their clients but write their own texts themselves. They sold others on the idea of no longer writing texts themselves but didn’t trust AI with their own stuff. AI book authors, myself included, describe AI-generated texts as high quality and write their books by hand. When your own name is on it, the quality apparently isn’t enough. This was a central contradiction of an entire industry. For me it was no longer a coincidence, I had seen it too often for that.

There are two possibilities. Either AI-generated content is high quality, in which case you could write your own stuff with it too. Or it isn’t, in which case you shouldn’t market it that way. My suspicion was confirmed: it primarily depends on who the text is for.

Because nobody talks about this, or nobody dared to, or nobody could really measure it. AI content is good enough for you but not for me. The experts use AI as support. What they publish themselves, they write differently and mostly by hand.

It just doesn’t stand out because you can’t check it. The recommendation is sold complete with expensive training and software you first have to set up. But what the consultant actually does, you never see.

In thirty years of consulting I’ve developed a rule of thumb: look at what someone does for themselves, not what they recommend to you. Anyone who recommends AI content and writes their own texts by hand says clearly enough what they really think.

What the recommendations and guides really mean is that AI texts are good enough for the masses. For the daily flood of content nobody really reads. For the newsletter you skim at best, or the blog post that stops being interesting after three sentences, it’s enough.

But when your name is on it and it’s supposed to really matter, it’s no longer enough and everyone knows it. These synthetic-sounding and extremely clever formulations become unreadable mass-produced text that doesn’t get beyond alphabet soup.

There are texts where the origin doesn’t matter. Forms, automated notifications or other standard texts, because nobody expects a human voice there. But when tools are recommended for marketing texts and customer communication, texts that are supposed to build trust, and those tools aren’t good enough for the recommender’s own work, then in my eyes that’s a lie. You’re selling a lot of efficiency at miserable quality.

AI-generated texts are generally presented as high quality. But these analysts would rather write their own texts by hand. I do it too. By now most authors have noticed the contradiction, yet what strikes me is that nobody really wants to talk about it, because everyone is still testing and analyzing and simply isn’t sure. I’m no smarter than other authors, but for me writing has a dimension of reflection, learning and joy in expression, which is why I can keep it fairly clearly apart. What I miss is the open discussion about it. It’s only ever about the effectiveness of content production, and a text gets reduced to that dimension alone. I wish for more resilience in genuinely hand-written work. I see it like high-priced, handcrafted products, watches or furniture, and I believe the value of hand-written texts will rise.