Why AI detectors can’t solve the problem they were built for
About a year and a half ago, I began deliberately deleting em dashes from my writing. It was for exactly the reason you’re thinking—because AI chatbots tend to overuse them; their presence was becoming a tell. Even if the words were 100% human-generated, I thought anyone reading might raise an eyebrow if they saw them and wonder, Is this AI? You can’t blame the world for its suspicion.…
For the past year and a half, I made a deliberate effort to remove em dashes from my writing, primarily because AI chatbots often overuse them, making their presence an obvious tell. It's clear that AI-generated content is prevalent online, making up half of all published articles, according to a Graphite analysis. While AI text doesn't automatically equate to poor quality, most people use it as a proxy for value or worth their time.
Consequently, AI detectors have gained widespread attention as a potential solution to this issue, but they've failed to address the problem for three main reasons.
Firstly, these detectors are unreliable, frequently yielding false positives. Secondly, even when a human-generated tell exists, it still plays a role in the reader's perception, which no software can detect. Lastly, there's a lack of consensus on what the actual problem entails. This third point became apparent during the recent case of billionaire Stanley Druckenmiller, who admitted to using AI to write a guest column for The Wall Street Journal.
Druckenmiller defended his use of AI, stating that he employs the technology for similar reasons he uses a calculator for math problems—because he's not a gifted writer and outsources the task. The editor of The Journal agreed that AI is an integral part of modern life, and people will continue using it to assist in their writing, whether for research, grammar checks, or editing.
However, the editorial page editor emphasized the importance of ensuring that the published content reflects the original argument and credibility of the author.
In response to this incident, Semafor conducted an analysis of the frequency of AI writing from op-ed contributors at The Journal, The New York Times, and The Washington Post. The results revealed that only about 3% of the analyzed articles contained entirely AI-generated content, with The Journal's proportion being slightly higher at 5%.
This relatively low rate has led to an outsized reaction, as people tend to question the effort and credibility of a piece once they suspect AI involvement. The concern stems from the belief that an AI-written article may indicate a lack of effort or thought, effectively turning style into a proxy for effort and credibility.
Written by urgent.news from Fast Company's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
