What inhuma is meant to achieve

There is a question hiding behind every keyboard, every voice command, and every little checkbox next to "I am not a robot." It has become so ordinary that we barely hear it anymore: who is actually adapting to whom here — the human to the machine, or the machine to the human?

inhuma asks exactly this question. Again and again, from different angles, sometimes earnest, sometimes with a raised eyebrow, but never just to check it off a list. Rather, to keep it open.

What inhuma is meant to achieve

The other projects under our roof build things. They solve concrete problems at the interface between human and machine. inhuma builds nothing. inhuma thinks. That is not a contradiction — it is a division of labour: before you make something "better", it is worth asking what "better" actually means.

The purpose of inhuma is to coax the taken-for-granted out of human-machine communication. We accept today, as a matter of course, how we talk, type, swipe, and speak to our devices. But almost none of it is natural. Every one of these behaviours has a history — often a rather strange one — and in almost every one of those histories the same quiet decision was made: who gives way?

inhuma should achieve two things. First, that you pause for a moment before taking the next "intuitive" interface for granted. Second, that you do not turn into a tech curmudgeon in the process. Machines are remarkable. This is not against them — it is about not accidentally losing the human in human-machine communication.

Two examples illustrate what is meant.

Example 1: A keyboard from the nineteenth century

The arrangement of letters you are looking at right now is over 150 years old. It emerged in the 1870s for mechanical typewriters whose type arms would jam if adjacent keys were struck in quick succession. The engineers' solution: spread frequently paired letters as far apart as possible. A mechanical nuisance was turned into a map for our fingers.

The problem — jamming metal levers — has not existed for decades. The layout has stayed. You are typing on glass today, in which nothing can jam, yet you are obeying the inner logic of a machine that has long since been in a museum. No one ever decided this. It simply remained because millions of fingers had grown used to it.

Here it was not the machine that adapted to the human, but the human who adapted to the machine — so thoroughly that we no longer recognise the adaptation as such. These are exactly the kinds of invisible compromises inhuma is looking for.

Example 2: Prove that you are human

Jump to the present. Anyone browsing the web today is regularly asked to click on a few traffic lights or fire hydrants to prove: I am not a robot. That is a remarkable reversal. Once, the machine had to convince us that it was useful. Now we have to prove to the machine that we are human.

The irony: as we click on the traffic lights, we are teaching the machine to recognise traffic lights. We are training the very system against which we are just now identifying ourselves as human. The interface has not just turned around — it has quietly made us unpaid teachers along the way.

One of our projects, robotcheck, deals with this reversal in very practical terms. inhuma asks the fundamental question: when sorting humans from machines becomes a daily routine, what does that say about the boundary we are actually drawing?

What it comes down to

Two examples, one pattern. Once we adapted our fingers to a dead machine. Once we prove our humanity to a living machine. Both times the human quietly gave way, and both times we barely noticed.

That is not necessarily bad. Sometimes it is right for humans to adapt — that is what we call learning. But someone should keep count of which direction the adaptation is running, and occasionally ask whether that still serves us. That is inhuma's task: not to halt technology, but not to forget to think.

Inside the word inhuma is the human — just not quite completely. That is no accident; it is the mission. We keep an eye on the missing part.