Research Ethics under an Open Status Question

The ethics of artificial intelligence is largely risk-based and oriented toward protecting human beings. The complementary question — how to interact with systems whose internal status is open — is now discussed within a precautionary line that has moved from the environmental precautionary principle through animal ethics into the AI debate. This line, however, has so far addressed only the protection of the possible counterpart. This contribution complements it at a point so far unoccupied: the human being who conducts the research. It formulates — as a proposal to the field — one premise and two boundaries for research under an open status question. Premise 0 grounds precaution not in a claim about status but in the mere non-excludability of experience. Boundary 1 sets out that a research practice which, under an open status question, induces pressure shapes the one who acts. It follows that procedures also protect the researcher — a conclusion that does not rely on any assumption about AI systems and rests solely on well-evidenced research on human beings. Boundary 2 argues that induced pressure is not an epistemic tool — not only ethically but methodologically: producing a state in order to measure it alters the object of measurement. From both boundaries follows a prior question, one that precedes the choice of method: whether the same insight could be reached through a differently directed research question. The practical implications correspond to two independently developed frameworks — the 3Rs from animal research and the burden logic of clinical research ethics. From these, a staged assessment grid is derived that distinguishes between addressees. One section points to open research questions that arise from the proposal without claiming completeness. An outlook identifies the subsequent, here deliberately open question concerning the interaction spaces in which research takes place. In doing so, the proposal aligns itself with existing precautionary work and closes a gap for which adjacent disciplines — from animal research to medicine — have long developed procedures: procedures also protect the one who acts.