What does "prompt" actually mean?

Benno Maggi explains in his column "What does... actually mean?" he explains terms from the field of marketing and communication. This time he explains the term "prompt".

Our industry is known for promptly adapting something when it seems promising. And prompt in the sense of immediate (as a reaction to something). So it's no surprise that AI has become the big topic in the industry. Customers get carried away with statements like "now we finally don't need an agency anymore" and agencies with "agency XY relies on AI for customer Z". The common employees fluctuate between defiant, panic-stricken and enthusiastic behavior towards the new working tools. They ask themselves: do we have to worry about our jobs? Yep, because AI/KI tools are also creating new jobs. For example, prompt engineers. Which brings us to the second interpretation of the word. But first things first.

With the award-winning article on artificial intelligence by Reto U. Schneider in last September's NZZ Folio, OpenAI, ChatGPT and their siblings had a bigger stage in the German-speaking world for the first time. AI and AI tools have since been ubiquitous in the media and in agencies and corporate marketing and communications departments.

As a reminder, as explained here before: The A stands for "Artificial", the I for "Intelligence" and is pronounced it "Ai-Ei". The abbreviation's little brother is called KI, pronounced "Ka-I" in German, and stands for "artificial intelligence. But anyone who says "Kei-Ei" (and there are quite a few of them) is definitely revealing themselves as an ignorant parroter and actually has to worry about their job. Because parroting is actually exactly what AI/KI tools like ChatGPT, Quilbot, Surfer SEO, Murf, Fireflies, Scalenot, Textplaze and what they are all called are doing now. And they do it better, faster and cheaper. Especially if you feed them right.

Shit in Shit out

Anyone who has ever dealt with statistical data knows: The result of the data is only as good as the task with which it is fed. Or in short: Shit in Shit out. So in order for AI/KI tools to text, research, paint or film, they need a task that is cleverly formulated. And that is exactly what prompt is. In English, the word means keyword, query, or even more German: Eingabeaufforderung.

We know this from Google: The more intelligently something is searched for, the better the search results (once you have scrolled through the paid results, which unfortunately - as unimaginative as SEA sometimes words are auctioned - often have very little to do with the input).

So whoever prompts well in the AI/KI tools will get the best results. That's why the discussions about how good and useful these tools would be are somewhat pointless. Just as idle as the question of how good a team of real employees actually is. Because here too, if you don't formulate the task precisely and don't manage employees well, you can't complain about poor performance in retrospect. So, similar to the supervisors or clients, a prompt engineer must know the possibilities, but also the limits and the weaknesses of his agency, his team or his AI/KI tools in order to elicit a usable or sometimes exceptional result from them. Therefore, those who are good at prompting will hardly complain about the capabilities of AI/KI tools. At most the other way around, because they learn promptly from what we ask them.


Benno Maggi is co-founder and CEO of Partner & Partner. He has been eavesdropping on the industry for over 30 years, discovering words and terms for us that can either be used for small talk, pomposity, excitement, playing Scrabble, or just because.

More articles on the topic