Deliberate imperfection and the effort heuristic

The advice is already circulating in professional forums: add typos to your text so it doesn’t look AI-generated. What does it say about us that imperfection has become a certificate of authenticity?

The logic behind it is less absurd than it sounds. Research by Schilke and Reimann (2025) showed that people who disclose their use of AI are systematically perceived as less trustworthy ¹. A 2024 study found that participants rated AI-generated ads as more attractive than human-made ones — until they were told the source. Once labeled “made by AI,” they were perceived as less authentic, less trustworthy, and with less emotional resonance ². The “AI” label acting as a switch that shuts off the receiver’s empathic connection.

The market’s response has been as ingenious as it is revealing. An entire industry of AI “humanizers” has emerged — tools like Undetectable AI, HIX Bypass, or Ryne AI — that promise to transform AI-generated text into content that passes as 100% human ³. And in professional forums, one of the most discussed proposals is the deliberate introduction of human errors, based on the premise that AI is “too perfect” and that humans are intrinsically erratic.

Think about that. A professional deliberately degrading an excellent result to make it seem “human enough.” Like a chef with the best oven on the market who swears he cooked over wood, because he knows you’d pay more for the smoke than for the flavor.

And here’s the paradox: it doesn’t even work. Adding random spelling errors can actually increase the probability of detection by 23%, since modern detectors don’t just look for correctness — they look for deep linguistic patterns and semantic fingerprints ³.

The underlying problem is more revealing than the tactic. What’s operating beneath the surface is what cognitive psychology calls the “effort heuristic”: a cognitive shortcut by which people estimate the quality or value of something based on the perceived effort, time, or difficulty that went into making it, rather than evaluating only its objective result ⁴. A client may be willing to pay a premium for a design that took weeks of manual drafts, but feel cheated if the same result is produced instantly through a prompt. Same pixel, same color, same emotion. Different price.

And here comes the uncomfortable question. If the result is objectively better, what are we protecting by hiding the tool? The quality of the work, or the illusion that value can only be born from sacrifice? As long as we keep measuring talent with pre-industrial metrics — time and sweat — we will keep incentivizing concealment. And deliberate imperfection will be the most absurd certificate of authenticity in our professional history.