How not to detect AI plagiarism

There are times when AI detection software is the best solution, but it is also worth exploring how to avoid situations where its use is necessary.

Rethink learning; rethink evaluation

• ask students to engage in assignments at which large language models (LLMs) are less effective, such as writing creative, opinionated, unexpected writing that expresses the students natural tone and voice

• instill students with the desire to learn, be creative, and master the material for its own sake (intrinsic motivation, passion)

• oral presentations, discussions, performances in the classroom where students are called on to act from memory or spontaneously will also help them develop useful skills for interacting with other humans (Lim, 2022)

• have students write assignments in class, on paper, or with access to LLMs prohibited

Incorporate AI as a tool in assignments

If you can't beat them, have them join us:

• Define a clear role for AI to play in an assignment, while delineating where the student should produce their own content. For example, students could critique the bias or inaccuracies of AI generated content, or experiment with alternative prompts and the output they generate.

Have a specific assignment which is AI-enabled, for example:

• Have students write up responses to an AI-generated critique of their essay, or write a plan to improve their weaknesses.

• Have students analyse biases in AI output and discuss the possible societal impacts this might have.

• Have students critique the use of AI in other contexts or by other users.

• Have students see if they can detect the differences between human- and AI-generated texts/ images.

• Let students use AI to learn a new skill, such as programming. Rather than seeing AI as a tool for avoiding work, they will be using it as a tool for learning, and will inevitably in the process see where it makes errors.