When humans must prove they are human – the case for human intelligence

The new AI age is bringing with it a peculiar irony: the machines built on the foundations of human creativity may soon render that very creativity as false, not by replacing it but by absorbing it so completely that AI copy becomes indistinguishable from human-generated content. 

The doomsday scenario will not come in the white heat of dramatic confrontation, but in a quieter, more relentless way as the novelist, journalist, copywriter or poet who writes with genuine feeling and competency is identified by an AI detector as a machine.

What is the alternative?

The pressure an unjust AI detection verdict brings can be very real. Academic institutions pass student work through AI detectors, sometimes with life-altering consequences. Publishers and editors, fearful of reputational harm, follow the same path. A writer who ‘fails’ the detection test faces a stark choice: either dispute the result or quietly adapt their style to raise fewer concerns.

This is probably one of the most dispiriting surrenders in the history of human communication: the deliberate introduction of error, clumsiness, and prosaic imperfection in order to appear so flawed that a machine will generously grant a certificate of humanity. The irony is that to please the algorithm, good writers must reduce their competence to be believed. As AI detectors accept these patterns, a race-to-the-bottom is triggered with human clumsiness increasing to avoid learned AI behaviour and the pointing digital digit.

Stubbornness may be the only strategy

Human writing is not valuable because it follows particular syntactic patterns. It is valuable because of the things that underpin it: experience, curiosity, obsession, irony, grief, or the sheer joy of language for its own sake. These are attributes that stretch far beyond sentence structure. They manifest themselves in the things a writer chooses to notice, the things an author chooses to ignore, and the paths a creator will choose to wander when granted the freedom to do so. What differentiates the human writer from AI are the idiosyncratic references, the surprising direction changes, and the humour that comes from lived experiences.

Arguably, the real winners in this doomsday scenario may be those who resist the pressure to contort themselves to satisfy algorithms (which might change rapidly). They will write with the confidence and stubbornness that allows them to be themselves, without apology, and trust that their work will find an audience appreciative of more than linguistic patterns. After all, the history of literature is marked by writers who were misread, misclassified or rejected by guardians proven wrong by the passage of time. The AI detector is simply a particularly obtuse guardian.

Trust is hard-won, and easily lost

The bigger issue is institutional. AI detectors are blunt instruments used to perform delicate surgery. When a university fails a student, or an editor rejects a submission on the basis of a probability score, they are not detecting AI. They are detecting conformity to patterns that represent the best of human creativity. The two are not the same.

In my opinion, the solution is not for human writers to abandon the skills that make them experts at their craft. The solution is for institutions to abandon their lazy faith in AI detectors and to turn instead to the human ability to exercise editorial judgement. Did I enjoy reading this piece? How did it make me feel as a reader? Machines can flag and accuse, but only humans can really judge.