Why Ethics Always Comes Last

Ethics always comes last. Especially when it comes to AI, ethics feels like a nuisance because it slows down the hunger for efficiency. So it ends up at the bottom of the list, after everything else has been checked off.

At some point the decision was made to include the ethics topic. Or it was forgotten and only noticed afterwards. In a way it is also a decision to just put something forgotten at the end of the list. And the moment someone decides the order of topics and ethics has to get back in line, that says something about relevance and priority.

At the beginning the excitement is huge about what AI can do and why it needs to be implemented now. There is discussion about the application, about the necessary tools and ideally the whole thing gets called Agentic AI and every department gets to marvel. Everyone looks at the results, what others have achieved and that the future is going to be magnificent. And then at the very end: ethics as a cautionary note that strict attention must be paid. Once more from the top: first the requirements, then the software development, tool implementation, initial tests, processes are rehearsed and then comes the conscience.

I am not talking about theory. I have seen these processes myself. Wherever I was involved, ethics never stood at the beginning of the discussion. Just like in pharmaceutical development it is first about the effect and then about the side effects. Financial products first paint the returns in the most lavish colors and then comes a risk disclaimer. This sequence is functional and established because whoever is excited right away scrutinizes less.

An AI strategy that started with ethics would raise different questions. What happens to the people whose work the machine takes over? Who owns the data? What does it mean when decisions about people are made by systems nobody fully understands?

That would require a different strategy, one that naturally slows many things down and would be considerably slower and more expensive to implement. It destroys an entire project euphoria when you don’t start with a rosy promise but with a complicated question about a problem nobody wants. But it would be an honest beginning.

The position of a topic in a document often defines its weight. What comes first is important by default and then frames everything that follows. What comes last is usually not even read anymore. Opinions are already formed and decisions already made. Ethics is never the main course but at best a decorative side dish that arrives cold on the plate.

The ethics sections usually read quite nicely and they do name the right topics. Wrapped in well-chosen formulations they present the subject seriously and in a way that is perfectly comprehensible for everyone. At least on the textual level.

But the argumentative structure contradicts the text as such because if ethics does not open the door at the entrance, it no longer has a real message beyond the idea that it is enough to think about the consequences at the end.

And that is exactly how it plays out in many companies. AI gets introduced, processes restructured, and based on the results it is optimized further and further. At some point, usually after something has gone wrong or a penalty from some higher court is looming, the paragraph on the ethics question gets read. I have also seen it added after the fact to appease courts. I have been told that happens quite frequently. And then committees are formed and guidelines written. All after the fact and with great seriousness. But only after the reality has long been established.

Many AI discussions reproduce this sequence instead of questioning it. One could show that ethics does not belong at the end because it is difficult but at the beginning because it is important. But just like in pharmaceuticals or finance, nobody likes to take off with the handbrake pulled.

The people involved would know better but the collectively agreed-upon silence is already a well-established system.

The result of an argument should be a clear statement. And the statement here is: Ethics is what is left over when everything else has been implemented.