Hsiao-Ping Hsu
9 min
This study investigates how pre-service primary teachers in Ireland engage with generative artificial intelligence (GenAI) for personal, academic, and lesson planning purposes. As initial practitioners, these teachers are at a critical stage where digital habits may influence future professional practices. The research seeks to understand if familiarity with GenAI in non-professional contexts translates into more effective or positive attitudes toward integrating these tools into formal classroom lesson planning.
Drawing on data from 100 student teachers at an Irish institution, the research employed a quantitative questionnaire survey. The study assessed three primary areas of GenAI engagement: personal use (e.g., entertainment, hobbies), academic use (e.g., writing assistance, project work), and lesson planning. Using descriptive statistics and Pearson correlation, the researcher examined how different types of prior experience shape perceptions of opportunities, challenges, ethical concerns, and the perceived need for professional development.
This study provides evidence that mere exposure to GenAI does not equate to pedagogical preparedness. Because academic use of GenAI can lead to both increased optimism and a potential underestimation of risks, teacher education programs should tailor their professional development. Rather than generic training, they should offer advanced paths for those already familiar with academic AI, specifically targeting critical evaluation and ethical considerations, while providing foundational training for those with only personal exposure.
Sam: Who actually answered? That matters if we're deciding whether these results describe student teachers generally.
Alex: The researchers invited students across the institution's primary teacher-education programmes through email. After excluding incomplete or careless responses, they analysed 100 questionnaires. Participation was voluntary and anonymous, and the participants formed a convenience sample, recruited from an accessible group.
Sam: So they sampled one institution, and students chose whether to respond. What did experience look like within that group?
Alex: Sixty-eight respondents reported having used a generative AI tool. ChatGPT dominated, though some students used other tools as well. Self-rated experience was limited overall, and lowest for lesson planning rather than personal or academic use.
Sam: Were they using AI to write complete lessons, or mainly to help with particular tasks?
Alex: The reported uses lean toward particular tasks. Brainstorming teaching or learning activities was the most common lesson-planning application, reported by 44 respondents. Developing teaching materials also featured prominently, while assessment and feedback were less common.
Sam: That sounds more like gathering ingredients than handing over the whole planning process. Did their attitudes reflect that distinction?
Alex: They saw efficiency as a benefit but were less convinced about the output itself. Agreement that AI could save planning time averaged about four out of five. Confidence in the accuracy and reliability of generated plans averaged about two and a half on that same scale.
Sam: So they could believe it saves time without believing it produces trustworthy lessons. What worried them most?
Alex: Over-reliance was a prominent concern. Students also worried about possible negative effects on children's learning. Those are perceptions, though: this study did not measure lesson quality or children's outcomes.
Sam: And ethics can mean several different things here. Which concerns did the researchers actually ask about?
Alex: They asked about educational inequity, data privacy, disclosure during teaching placements, and whether unauthorised use counts as cheating. Responses on those issues were mixed. Students leaned toward supporting disclosure, while agreement that unauthorised use was cheating was lower.
Sam: That makes guidance relevant even before anyone decides whether the tools are effective. What happened when they connected these perceptions to prior use?
Alex: This is the central comparison. Personal AI experience showed no statistically significant relationship with the lesson-planning perception measures. Academic experience was associated with more perceived opportunities and a greater desire for training in effective and ethical use.
Sam: Does the absence of a significant relationship mean personal experience has no relevance?
Alex: No. It means this sample did not provide evidence of those relationships. The result challenges the assumption that personal familiarity is enough, but it doesn't establish that personal experience can never matter.
Sam: The academic pattern sounds encouraging for training. Was there a less reassuring side?
Alex: Academic experience was also associated with fewer perceived challenges and fewer ethical concerns. Its relationship with perceived opportunities was the strongest of those reported associations. The authors suggest familiarity might create blind spots, even while students recognise a need for guidance.
Sam: But students who already see fewer risks might be more willing to use AI for academic work. Can this survey distinguish that from experience changing their views?
Alex: It cannot. These are correlations between self-reports, not evidence that academic use caused optimism or reduced concern. The proposed blind spots are an interpretation, not a demonstrated failure to identify ethical problems.
Sam: With that boundary in place, what do the authors recommend teacher educators actually do?
Alex: They propose foundational training for students whose experience is mainly personal, potentially similar to training for newcomers. For students with academic experience, they propose more advanced sessions that build on existing uses. Both routes should address practical challenges and ethics.
Sam: What would building on academic experience look like, rather than just giving experienced users a harder tutorial?
Alex: The authors give a concrete example. A student who uses AI to generate assignment ideas could learn to generate teaching ideas for topics children find abstract or challenging. That's a proposed training activity, not an intervention evaluated in this survey.
Sam: Beyond causality, what's the biggest reason not to turn those recommendations into a general rule?
Alex: The sample came from a single institution, through voluntary convenience sampling. The authors acknowledge potential sampling bias and limited diversity, which restrict generalisation. Use was also at an early stage, so perceptions may reflect limited exposure.
Sam: Then who should give the full document their time, and where should they begin?
Alex: Teacher educators designing AI training, and researchers studying educational technology adoption. Start with the results comparing personal and academic experience, then read the limitations in the conclusion. The discussion offers training proposals, but those still need evaluation.
Sam: And for someone who mainly wants the finding to carry into a conversation with colleagues?
Alex: In this sample, academic AI experience was associated with seeing more promise in lesson planning; personal experience wasn't. The practical line to remember is: familiarity with AI outside the classroom is not evidence of readiness inside it.
Sam: Thanks for listening.