Can One AI Unmask Another? Pangram, an Almost Infallible Tool for Detecting Machine-Created Texts and Novels
6 Articles
6 Articles
The American start-up, which claims it can unmask texts written by artificial intelligence with a very high degree of confidence, has found itself at the center of the controversy surrounding Haitian-Québécois writer Thélyson Orélien and his breakthrough novel 'C'était C'était ou Mourir.'
To test the artificial intelligence detector against Thélyson Orélien, we submitted Peguy, Zola, Dostoevski, Hugo and even the Declaration of the Rights of Man and the Citizen. With scores reaching 50%, the dead come to disturb the trial made during the lifetime.
The rise of generational artificial intelligence (AI), capable of producing texts similar to human writing, has led to the creation of multiple tools designed to detect whether a writing was produced using this technology. However, the reliability of these systems continues to be the subject of analysis. In France, one of the literary phenomena of [...]
At the heart of the controversy about the novelist Thélyson Orélien accused of having written his book with the AI, the start-up Pangram ensures that it is possible to unmask with great reliability the texts written by an artificial intelligence.
We have submitted more than a hundred texts to him. Pangram almost always recognizes humans and machines, but is (easily) berner.
Coverage Details
Bias Distribution
- 100% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
