Should We Let Chatbots Walk Away With Literary Prizes?

Ringing the death knell for human writing

Photo by Mohamed Nohassi on Unsplash

“The age when you can run a literary magazine on trust is over. You can either decide AI submissions are fine, or you can filter for them.” — Kelsey Piper, The Argument

Literary magazines have a tough job on their hands: to distinguish human-written content from AI-authored stories.

The available AI detection tools are inefficient. They can’t catch AI-generated content with an acceptable degree of accuracy. To add to the problem, we have AI tools that “humanise” AI- content.

A literary competition tests human creativity, especially the storytelling ability. Literary magazines or organisations that conduct competitions are supposed to ensure that the winners are humans, and not chatbots.

Kelsey Piper says that many winning stories in the recent literary competition held by the Commonwealth Foundation appeared to have been written by AI. The organisation did not screen for AI-generated content. It merely relied on a declaration from the authors that they have not used AI to write their stories.

Kelsey Piper says that we should do two things to detect AI-produced content. Use expert human judgment or the best available AI detection tool.

Experts who have read a lot of AI-generated content can easily find out whether a story is human-written or machine-generated. For example, many winning stories contained metpahors that no human writer would use. For example, “Her laugh is as bright as zinc.”

According to Kelsey Piper, there’s a tool called Pangram that has proved to be fairly accurate in detecting AI-written content.Pangram matched the accuracy of expert human AI-detectors. It also had fewer false positives. That is, it did not falsely accuse human writers of using AI.

The Commonwealth Foundation said,

“The prize has an important role in supporting underserved communities and enabling new writers to find a voice.”

A laudable objective, no doubt. That doesn’t necessarily mean we can bypass AI-detection.

Relying on the writers’ self-certification of full authorship seems naive because trust is earned and demonstrated beyond doubt, not merely assumed.

Letting chatbots walk away with literary prizes is allowing the human citadel of authentic writing fall without a fight.

Thanks for reading!

Source