Not long ago it was: just ask Reiner, he knows his way around artificial intelligence, he’ll sort you out. These days it’s no longer “ask Reiner” but “ask Claude”. Claude knows everything.
And what happens? People who know less than I do, and far less than Claude, stop asking me and ask the all-knowing Claude instead.
That’s when the alarm bells go off for me. Because of course Claude always has an answer. And it’s probably different from mine. My answer isn’t a technical solution - it’s a set of counter-questions. (Where I grew up, in the Rhineland, we’re famous for answering a question with a question. Ask a Rhinelander what the weather will do and you’ll get: “What the weather will do?” - followed straight away by the answer: “Good, of course!” We always need a few extra words.) Above all, though, mine are questions you should ask yourself first, before you ever ask Claude.
Which means: the AI experts who got the whole company to where it stands now will be the first ones eaten. Because the people deciding - who aren’t that far along themselves - don’t grasp how important it is to have someone desperately trying to keep the overview.
The great levelling machine
Interestingly, that’s why I see more and more of the pioneers leaving the field. They notice a great, powerful levelling machine flattening everything, leaving almost no room to stand out. Everyone turns up with some solution or other now. And it’s incredibly tiring, almost impossible, to compare two complex, complicated AI solutions against each other.
You could of course use AI for that too. But when my colleague flings something into the machine without much thought and gets back something wonderfully shiny that even seems to run — then as the AI expert I don’t need to stand in his way. I can just say: right, try getting it into production. Off your computer, onto a server. And make sure it works not only now but next week too. Then it looks rather different.
As an AI expert in marketing, though, you can barely score points anymore. Sure, you could hop from company to company to bring them up to speed too. But every client you soon leave again is one client fewer. And in the time it takes, the others have long got there themselves. So the differentiation has to happen somewhere else. But where?
Originally published auf Deutsch at reinergaertner.de (est. 1997, older than Google). AI helped translate this. I helped introduce the errors.