I Am Not a Robot

Kevin Purkhauser, Karen Lancaster

Keywords: Inverted Turing Test, CAPTCHA, human-machine distinction, productive imperfection, responsibility, large language models

TL;DR: Humans now have to prove they are not machines, and any test designed to verify this is structurally temporary. The problem is the ceiling-based logic underlying such tests.

Abstract: Seventy-five years after Turing introduced the Imitation Game, public discourse increasingly treats contemporary AI, especially large language models, as having “passed“ his test. This paper argues that the more urgent question has quietly inverted: can humans still demonstrate that they are not machines. We call this the Inverse Turing Test (ITT). We trace this inversion through the rise and collapse of CAPTCHA-style systems, which asked humans to solve tasks once difficult for machines but systematically failed because they measured outputs rather than processes and imposed a performance ceiling that machines eventually reached. We argue that a viable ITT must instead be organised around a floor of humanness rather than a ceiling of machine performance. Drawing on four candidate criteria -- embodied situatedness, temporal depth, social responsibility, and productive imperfection -- we characterise humanness as an ontological condition as well as a broad, inconsistent, context-dependent range of behaviour that eludes purely output-based tests. On this view, what counts as “human“ or “machine“ is not a fixed boundary but a moving threshold, continually renegotiated as capabilities, contexts, and criteria change.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 3