Even though AI-generated videos may currently cause some unease among active filmmakers, it is becoming clear that future developments cannot be reliably predicted. A particularly good example is the relatively young job description of the Prompt Engineer.
About 18 months ago, one could read everywhere that this was supposed to be a profession with excellent future prospects, since Prompt Engineers would ultimately be the managers of the emerging AI systems. In the best hype manner, it was concluded that in the medium term, every Prompt Engineer would displace other professions that can be filled by AI models. Copywriters? Cameramen? Cancer diagnosticians? Really? In the future, the Prompt Engineer could take over everything as the tamer of the AI models.
The job description of the AI Prompt Engineer has already disappeared
But now the opposite has happened. As the Wall Street Journal reports, the term is practically no longer found in current job descriptions from 2025. Which is remarkable, since 18 months ago it was assumed that such positions should be worth up to ,000. A survey by Microsoft also shows that the term Prompt Engineering now only appears at the very end of the list of new AI professions.
But what happened? First of all, there is no well-founded knowledge base for this profession. Whether someone is a good Prompt Engineer can ultimately only be measured very individually by their results. At the same time, special prompting know-how is relatively short-lived. What works well with an AI model today can lead to completely different results tomorrow.
In our experience, however, the crucial point is likely to be quite different: Good prompts still require the relevant background knowledge on the subject. Practically speaking: You have to understand something about filmmaking to prompt a good AI film, just as you have to understand something about programming to have an AI write its own routines via prompt.
Even with AI, you remain in control of the output for the time being and can "only" fall back on more or less intelligent AI employees, which you can only manage effectively if you yourself understand something about the task.
This thesis is also supported by the fact that, according to the Wall Street Journal, large companies have never even hired special Prompt Engineers. They do not see these skills as an independent profession, but rather as an additional qualification that can also be taught within the existing workforce through further training. And even their relevance seems to be tending to decline: According to the WSJ, modern AI systems are now achieving better results in dialogue with the user. And last but not least, immediate results now also allow you to simply try more until the result is as desired. The biggest obstacles to integration into many workflows are therefore AI anxieties, but almost never a lack of Prompt Engineers.
;)