Writers Face Job Losses Over Faulty AI Detectors
Image from Unsplash: Infralist.com, @infralist
AI detectors are flagging writers incorrectly, causing job losses. Despite AI’s rise, these false flags harm writers who depend on their skills. This issue calls for careful AI use to protect professionals.
Kimberly Gasuras, a journalist from Ohio, lost her job due to false AI flags. Despite writing herself, AI tools misidentified her work. She moved to freelance platforms but faced bans for incorrect AI detections.
Tools like Originality.AI and GPTZero boast high accuracy but often fail. Studies show frequent inaccuracies, leading to erroneous claims about AI use. Major universities have banned these tools due to their unreliability, yet companies still use them to curb AI-generated spam.
These detectors review texts for AI traits, but their methods produce false positives. Writers suffer job losses and financial strain as a result. Originality.AI and similar firms defend their products yet acknowledge flaws.
The industry remains divided. Experts recommend discussing AI writing’s social impacts over-relying on imperfect detection tools.
This kind of event highlights the importance of not relying too much on AI detectors, as they are not fail-safe. They might mark very well-written text as “non-human.”At Heartspace, we think that some AI detectors can provide value; they can help you improve your text and see where your writing becomes “stiff” and robot-like so that you can make it more vivid. At the same time, in some contexts, that stiff text is needed as expected in that particular setting. A high “AI percentage” should not be a reason to fire someone. For example, most of this text was written by me, by hand, but detected as AI anyway. Sometimes, I have had text written by our AI ghostwriter, Nirvana AI, that gets higher human ratings than I do. Still, I use detectors like GPTzero (which I think does a better job than others) to help me improve on text, as I do with Grammarly. I do not take everything they say as the truth but as input to improve the texts I write. AI detectors are just another tool, and they can have their purpose in some settings, but the conclusion remains:
You can’t trust AI detectors to detect AI, and I would never fire someone just because they get a high AI rating on a text.
For more details, see the article: AI Detectors Get It Wrong. Writers Are Being Fired Anyway from Gizmodo.