Turnitin AI writing detector

Turnitin's AI writing detection tool aims to help educators identify content potentially generated by AI writing tools such as ChatGPT. 

The tool analyses submitted text to detect patterns and characteristics typical of AI-generated content, and the breaks content into smaller chunks or sentences, which are then overlapped to capture their context. Each portion of text is given a score between 0 and 1 to determine if it was likely written by a human (0) or AI (1), with AI-generated text only flagged if it is 98 percent certain.

The tool provides an overall percentage of the document that may be AI-generated, and highlights specific portions of text likely to be AI-written, including potentially AI-paraphrased content.

Risks and harms 🛑

Turnitin's AI writing detection tool is seen to raise many concerns, including:

• False positives. The system may incorrectly flag legitimate student work as AI-generated, leading to unwarranted accusations of academic dishonesty. This can have serious consequences for students, including disciplinary actions and damage to their academic reputation.

• Loss of creativity. Students may feel discouraged from expressing their ideas freely if they fear their work will be misidentified as AI-generated. This could stifle creativity and critical thinking, as students might resort to more formulaic writing styles to avoid detection.

• Cultural diversity. Different writing styles, dialects, and cultural expressions may be misinterpreted by AI detection systems, leading to biased outcomes. This can disadvantage students from diverse backgrounds whose writing may not conform to conventional academic standards.

• Overreliance on technology. Educators may become overly reliant on Turnitin and other AI detection tools, potentially neglecting the importance of teaching writing skills and critical thinking. This could lead to a diminished focus on developing students' abilities to articulate their thoughts effectively.

• Learning culture. There is a risk that institutions may use AI detection tools as a punitive measure rather than as a means of fostering academic integrity. This could lead to a culture of fear rather than one of learning and growth.

• Equity and access. Not all students have the same access to resources or support for developing their writing skills. Those who struggle with writing may be disproportionately affected by AI detection tools, facing harsher scrutiny compared to their peers.

• Privacy. The use of AI detection tools may raise privacy issues, particularly if student work is stored or analyzed in ways that are not fully disclosed. Students may be uncomfortable with their writing being subjected to algorithmic scrutiny.

• Trust. The implementation of AI detection tools can create an atmosphere of suspicion between students and educators. If students feel they are being monitored for dishonesty, it may undermine the trust necessary for a positive learning environment.

Transparency and accountability 🙈

Turnitin's AI writing detection tool is seen to suffer from several transparency and accountability limitations:

• Black box algorithm. The exact workings of Turnitin's AI detection model are not publicly disclosed, making it difficult for users to understand how decisions are made.

• Explainability. The tool provides an AI probability percentage but does not offer detailed explanations for why specific text is flagged as AI-generated or paraphrased