4.2. Practical for the lecturer

The tasks involving the use of AI applications provide students with the opportunity to practice AI literacy by teaching them to evaluate citation and factual accuracy, identify bias in texts, and analyse the impact of style and tone in text editing.

Here are some sample tasks that can be used to develop AI literacy. By completing these tasks, students will have the opportunity to experiment with AI applications in a practical setting during their studies.

• Citation and References: Students will write an essay on a given topic using a generative AI application, including at least 20 references to recently published academic literature. They will then check the references — the sources must exist and align with what the AI-generated essay claims. If students are unable to locate the referenced source or discover a factual error, they must find the actual source or revise the essay. The list of used literature must include all the references generated by the AI application as well as any corrections made by the student.

• Fact-Checking: The instructor provides students with an AI-generated essay on a topic covered in the lecture/seminar, and students will verify the facts and references, annotating the essay as they go.

• Comparing Sources: Students will compare a primary source with both a human-generated and an AI-generated secondary source on that primary source.

• AI Bias: Students will generate an argument with AI and discuss the ways in which the generated text might be biased. Could the issue lie in the training dataset, or in how the dataset has been applied? Does rephrasing the prompt change the result?

• Editing with AI: Students will write a short text in class and ask the AI to edit it, creating different versions of the text: changes in tone and style, rewriting in literary or poetic language, using different analogies, metaphors, and figures of speech, or rewriting in the style of a sermon or a New York Times article.

Task

Create a task that develops AI usage skills within one of your courses.

How to test the suitability of your assessment methods in the age of AI?

Always test your assignment instructions with different text-generating AI applications before giving them to students and analyse the results they produce. AI-generated (independent) work often receives a grade of C. Instead of banning the use of AI applications, ban work at the C level — completely AI-generated solutions or essays should receive a grade of F going forward. Adjust your assessment criteria and matrices accordingly.

For example, refer to the essay evaluation matrix created by Bowen and Watson (2024).

Example of an essay evaluation matrix (adapted from Bowen and Watson, 2024).

	None (0%)	AI Essay (50%)	Good (75%)	Very Good (90-100%)
Thesis, analysis	No thesis.	The essay focuses on one idea (thesis). Examples are present but are general or unrelated to the paragraph’s main point.	Two-part thesis. Examples support the claims but are not strong or complete enough.	Three-part thesis. Many suitable examples for each claim. Each paragraph refers back to the introduction and thesis.
Structure	No structure.	The structure is clear, but some paragraphs should be moved. Transitions between paragraphs are not smooth. The introduction and conclusion need improvement or do not complement each other.	Each paragraph focuses on one idea/claim. Paragraphs are in a logical order. Transitions between paragraphs could be smoother. The introduction and/or conclusion do not complement each other.	The introduction and conclusion complement each other. Each paragraph focuses on one key idea. Paragraphs are in a logical order. Transitions between paragraphs are smooth.
Language	Frequent language errors make the text hard to understand.	No language errors, but sentences are repetitively structured and the text is monotonous.	No language errors, sentence structures are varied but could be clearer.	No language errors, sentences are varied in structure and clear.
Style and Author’s Voice	The author’s voice and audience awareness are missing.	The text is general, with the author’s voice not heard and no consideration of the audience.	The author’s voice is present, and the audience is generally considered.	The author’s voice is strong, and the audience is always considered.
References	References are missing.	References are present, but they are not enough; they are inaccurate or from unreliable sources.	References are present, properly formatted, and almost always from reliable sources.	References are present, properly formatted, and always from reliable sources.

Here are some recommendations if you think your evaluation method has become obsolete.

1. Reviewing and adapting assignments in the context of AI use

• Review how assignments support the achievement of learning outcomes. Define which AI usage might hinder the assignment’s purpose. Allow students to use AI applications if they help understand the process and outcome. Ask students to provide an overview of their use of AI and the lessons they learned.
• Create assignments that require students to ask the right questions and verify and assess the results of AI usage.
• Include questions based on the student’s opinion, experience, or analytical skills. Evaluate both the problem-solving process and the responses.
• Share experiences with colleagues on the use of AI applications and learn together.

2. Strategies for academic integrity and work validation

• Focus on objectives and integrity—use an integrity pledge where students confirm that they have completed the work themselves. Explain why it is essential to complete the task without outside help.
• Use alternative task forms and validation methods for grading, such as oral presentations, creative tasks, or paper-based tests.
• Create questions about the student’s unique experiences, interests, and background or non-public content, such as what was discussed or done in (video)lectures/seminars/labs.
• Use group work for assessment, where all group members respond orally and engage in discussions and peer feedback.

3. Limiting AI usage and supporting meaningful learning

• Plan tasks where it is challenging to use AI, such as case studies, experiments, observations, handling local issues, or novel sources. Explain the relationship between the process and the result in grading.
• Use written exams, (video)presentations, or timed in-class tests. For example, copy-pasting cannot be done in Moodle tests, but one could take a photo of text with their phone, ask for a response from an AI, and write it into the Moodle test. Time limits are an effective tool for preventing this.

4. Developing AI usage skills

• Teach students how to communicate effectively with AI applications and always ask them to justify the results they receive.
• Provide specific guidelines on how to cite AI usage (see also AI use in thesis writing).

5. Developing tasks and assessments in collaboration with AI

• Use grading rubrics available to students before they start the task.
• It is helpful to ask AI for suggestions on how to adjust your tasks and grading:
• “Suggest ten ways to make this task (economics/theatre studies, etc.) more interesting for students.”
• “How can students use an AI for this task?”
• “How could I adjust the task so that students would not use AI for it?”
• “How could students use AI to solve this task? What help do they need to figure out how to use AI effectively?”

Read more:

• TextGenEd: Teaching with Text Generation Technologies https://wac.colostate.edu/repository/collections/continuing-experiments/
• “TextGenEd” – A collection for educators on learning in the new era https://etu.ut.ee/ai/transforming-learning-exploring-textgened-and-the-new-era-of-text-generation-in-education/