Using an AI diary in journalism education: an experiment in encouraging transparency — and critical thinking
0 likes9 views
Presentation at the conference Journalism Education Reimagined: Global Perspectives, Future Pathways, hosted by Birmingham City University on July 8 2025. More info at https://onlinejournalismblog.com/2025/07/08/this-is-what-happened-when-i-asked-journalism-students-to-keep-an-ai-diary/
Using an AI diary in journalism education: an experiment in encouraging transparency — and critical thinking
- 1. Tell the truth. An experiment in encouraging transparency — and critical thinking
- 2. TL;DR ● I experimented with a new form of assessment. ● It worked. ● Mostly.
- 3. Journalism I&E* in 2024 ● 3rd year BA Journalism & Sports Journalism ● Data journalism module, 10 sessions ● Session 9: Using genAI to reflect and improve ● Getting the best marks (vs plagiarism) ● Employability: prompt design *Journalism Innovation and Entrepreneurship
- 4. GenAI tool use not mentioned If you’ve used GenAI tools without saying so, this is considered Category A plagiarism and a zero mark 0% ● Attribute any use of AI ● Make it clear what is your work, and what is not ● Include evidence to clarify ● Identify what you did well — and could do better Evaluation includes evidence E.g. “I used ChatGPT to generate ideas around my dataset (see Appendix A)” 52-8% ● Copy and paste the prompt(s) and response(s) in full ● Use a different appendix for each example ● Reflect on them Evidence and references E.g. “I used ChatGPT to generate ideas (see Appendix A) using techniques outlined by Marconi (2023)” 62-8% ● Draw on original sources used in lectures ● Focus on practical literature — don’t quote facts and stats ● Identify next steps + critical reflection and experimentation E.g. “I added prompts to guard against bias in terms of ethnicity, gender etc. (Heikkilä 2023. See Appendix B)” 72%+ ● Read about good practice and try those techniques ● Don’t settle for the first results: experiment ● Reflect on what works and what doesn’t Different levels of use/evaluation Note: grades indicate credit for use of genAI only, not the overall grade
- 5. View chart ‘story’ at https://public.flourish.studio/story/2594313/
- 6. Good, but not good enough ● Lack of transparency about editing ● Lack of use for review ● Students could reflect more critically ● Students reading to pass, not to learn ● Not having the ‘codes’ to access industry
- 7. What I did in 2025 ● Replace evaluation with AI diary ● AI in almost all classes ● Requirement for proof of interviews ● AI opt-out option ● Formative feedback in classes & via email ● CustomGPT ‘Paul’s mini editor’
- 9. “Students can articulate prompts that reflect their grasp of a subject, showcasing their comprehension.” Prompting Change: Exploring Prompt Engineering in Large Language Model AI and Its Potential to Transform Education (Cain 2023)
- 10. “Those who trust AI more tend to engage in less critical thinking, while those who have greater confidence in their own abilities are more likely to scrutinise AI-generated outputs” The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers
- 11. Week Lecture Workshop 1 Assessment/genAI Prompt design & AI diary (ideas) 2 Data-driven stories RAG, 'teaching' AI (7 angles) 3 Visualisation Writing about numbers/Datavis 4 Sorting and filtering Pivot tables 5 FOI Sending an FOI/ Using NotebookLM to dig into docs 7 - Diversity & inclusion: football club boards/Eid ideas + AI 8 Critical thinking & AI Interviewees: PEER + AI 9 Drafting data news stories Drafting data features 10 Reviewing with AI Story surgery/diary review 11 Assignment checklist Images, customGPTs
- 12. ● Improved transparency around editing ● Improved use for review ● Students reflected more critically ● New employable cross-disciplinary skillset Successes
- 13. https://public.flourish.studio/visualisation/23784406/ Li’s high level categories of journalism
- 14. “At the start of the module I was more naive and close-minded to the possibilities of AI and how useful it can be for journalists - particularly for idea development.”
- 15. “I used to be a fan of AI but after finding out how misleading it can be I work towards using it less.”
- 16. “I thought it was more reliable than it was and would believe the majority of what it would say. I didn't know that you could inject information into AI and get more out of AI through professional techniques.”
- 19. ● No word limit (one diary was 21,000 words!) ● Iterative rather than summative (+ feedback) ● A ‘moment of reflection’ made room for… ● Better planning and more reviewing ● Applied knowledge, not recited ● Employability and motivation? Why did it work?
- 20. End of module survey (n = 19)
- 21. End of module survey (n = 19)
- 22. What next? ● Require each story to indicate how AI was used ● Industry transparency practices (when/how) ● Teach editing AI-generated material ● Build confidence to correct AI quirks ● Tackle token references in diary (tasks? Tests?) ● A summary at the end of the diary
- 23. Templates, checklists? ● A risk assessment of genAI + measures taken ● A requirement to ask genAI for feedback ● “Breaking the illusion of explanatory depth” by choosing from multiple responses (Mollick & Mollick 2022)
- 24. Questions?