In the same way that Generative AI tools can create unreliable material, AI content detectors are not always accurate in their assessments. OpenAI, the creator of ChatGPT has said of their own detection tool: "We believe [the detection tool] does not have high enough accuracy for standalone detection and needs to be paired with metadata-based approaches, human judgment, and public education to be more effective." For more information on AI content detectors, check out Illinois State University's guide to AI content detectors.
Here are a few examples of AI Content Detectors:
- GPTZero: An AI detection tool developed by OpenAI to detect ChatGPT, GPT4, BARD, Llama, and other AI models.
- Turnitin: An AI detection tool designed for educators.
- WinstonAI: Detects text written by ChatGPT, GPT-4, Bard, Bing Chat, Claude, and many more Large Language Models.
- Content at Scale: Detects text written by ChatGPT, GPT-4, & Bard.
Check out these articles about AI Content Detectors:
- "The Trouble With AI Writing Detection," Inside Higher Education (Elizabeth Steere, 10/18/23)
- "Universities Rethink Using AI Writing Detectors to Vet Students’ Work," Blooomberh (Shirin Ghaffary, 9/21/23)
- "Our Obsession with Cheating is Ruining Our Relationship with Students," Substack (Marc Watkins, 1/6/23)
- "How Easy is it to Fool AI Detection Tools?" New York Times (Stuart A. Thompson and Tiffany Hsu, 6/28/23)
- "Turnitin’s AI Detector: Higher-Than-Expected False Positives," Inside Higher Ed (Susan D’Agostino, 6/1/23)
- "A college student created an app that can tell whether AI wrote an essay," NPR (Emma Bowman, 1/9/23)
- "A Plagiarism Detector Will Try to Catch Students Who Cheat With ChatGPT," Chronicle of Higher Education (Eva Surovell, 4/3/23)
- "Programs to detect AI discriminate against non-native English speakers, shows study," The Guardian (Ian Sample, 7/10/23)
