AI has no need to become creative

When we speak about artificial intelligence, the question regarding jobs comes to mind. How many jobs will artificial intelligence cost? The answer to this question is quite simple, nobody knows for sure. There are numerous studies on the subject, sometimes 20%, sometimes half, then frightening 80%. In the end, all that remains is the certainty that the new Colleague, Mr. Computer, is soon replacing some folk’s positions.

A few weeks ago, I read about a survey according to which most employees assume, that artificial intelligence cost jobs in their industry. However, the same majority believes that their jobs were not in danger. Only if you’re not mistaken…?  If anyone needs a job guarantee, my suggestion is, learn to become a craftsman. High-quality handicraft is becoming a prestige object to stand out from industrial mass production. Interestingly, one or two industrial workers maintain a relatively secure job, at least in those sectors in which automation is mostly exhausted.

Employees that do have to tremble are those working in offices. Making calculations, arranging appointments, sending out regular e-mails to customers – you’d ask yourself why our colleague Computer is not doing it already?

However, the supposedly creative ones amongst us are also at risk. Let’s take what I’m doing right now. I’m writing a text. Well, I might as well be an algorithm, the reader wouldn’t notice the difference. Even today, a program of this kind, for example, can take over the entire local sports reporting, and occasionally they already do so. Just as news agencies experiment with having their texts written by algorithms, not every product description in an online shop still has a human writer. Great literature will indeed still be the privilege of excellent writers, but the robot author can already write off-the-shelf novels.


I worked for a few years in an agency that also designed websites. I admit not to have an eye for design, but as a computer scientist, I have a particularly sharp eye for patterns. Moreover, that’s why I’ve seen a total of five website designs there in the two years, or so. Of course, they sold lots more, but most of them were according to two different patterns. Just a little different in color, here and there a different accent, but in principle always the same design structure. The web designers there probably didn’t even notice that for them every page was new. It is a little bit like an average novel. Their stories work according to a few specific patterns and with a handful of archetypes. Here these archetypes gather in a medieval scene, there on a foreign planet.

Perhaps we should dispel a prejudice at this point. The term “artificial intelligence” is somewhat misleading. The term “machine learning,” which is also in circulation, targets it much better. In simple terms, artificial intelligence works as follows: The programmer writes a self-learning algorithm. This one then tries A, realizes A is crap, so he tries B. If B is crap too, he tries C. C is good, but not good enough yet, why not choose Ca, is crap, so try Cb. That’s all right again, and so the game continues. In principle, machine learning is only the old “try and error” game, but thanks to the increased performance of the hardware, all this is happening much faster and more effectively today. For some outsiders, this can certainly give the impression of artificial intelligence. However, among us, even the Google program that recently beat the best Go player in the world is still as stupid as the Z1 that Konrad Zuse built in the middle of the last century. He’s just fast enough not to notice anyone.

So that also applies to the design. The design works according to specific laws. There is the golden ratio, and there are some “laws” for the combination of accurate colors, there are rules for the saturation of colors, and so on. An algorithm can learn and apply all this much better than a human designer. At some point, the algorithm has collected a critical number of elements that it only needs to combine user data about the customer, there it is, the individual design of the website. Website building kits like or Jimdo are already experimenting with these programs and it is only a matter of time before most customers ask themselves why they should pay expensive agency hourly rates if an algorithm does deliver the same result – perhaps even a better one, because it does not violate any of the unwritten design laws. Sounds unlikely? No, not really. We should not forget that people find such thing most beautiful, which is the least different from the average. Some supermodels may not like it, but they only have their model jobs because they are average looking.

Whether artificial intelligence someday stands up to real creativity, plays in the same category as the question of a possibly evolving consciousness. It’s highly speculative because even if it is possible to reproduce a human brain accurately, it does not automatically have to develop knowledge. However, the algorithm does not require any creativity for its actual purpose; the human designer, whom it replaces for cost reasons, is not creative either. It does what the algorithm does, follows design rules and recomposes familiar elements. Both create only the appearance of creativity, and both are probably not aware of that.

Real creativity, on the other hand, remains an isolated case, left to the few people who can create something creative. ’til now?