The research team employed a mixed-methods approach, a design philosophy that deliberately combines numerical measurement with narrative depth. Specifically, they used what methodologists call a sequential explanatory design, in which a quantitative phase is conducted first and its results then inform and are interpreted through a subsequent qualitative phase. In this study, the quantitative component took the form of a survey, while the qualitative component was structured as a case study. This sequencing matters: by first establishing how much anxiety exists across a broad group, the researchers could then use interviews to explore why those levels exist and what they mean to the individuals concerned, giving the numbers a human voice.

The quantitative phase drew on a population of 208 pre-service teachers enrolled in the science education department of the faculty of education at a state university in Turkey. From that population, the researchers selected a sample of 162 pre-service science teachers using simple random sampling, a technique in which every member of the population has an equal chance of being included. Random sampling of this kind is a cornerstone of survey methodology because it reduces the risk of systematic bias in who ends up answering the questionnaire, making the resulting descriptive and inferential statistics more defensible as a portrait of the wider group. The instrument administered to these 162 students was a scale designed to gauge levels of anxiety specifically about artificial intelligence applications.

When the scale data were analyzed using both descriptive statistics, which summarize the distribution of responses, and inferential statistics, which test whether observed patterns are likely to hold beyond the sample, the central finding emerged clearly: the pre-service science teachers reported a moderate level of artificial intelligence anxiety. That middle-of-the-road result is itself informative. It suggests that these future educators are not experiencing the acute alarm sometimes portrayed in popular discussions of AI, but neither are they indifferent or fully at ease. A moderate reading points to a population that is aware of the technology, uncertain about its implications, and receptive—potentially—to training that could tip the balance toward confidence.

To understand what lay behind that moderate score, the researchers turned to the qualitative phase of the sequential explanatory design. They conducted semi-structured interviews with eight pre-service science teachers, selected for maximum diversity, a purposive sampling strategy that deliberately seeks participants who differ from one another in relevant characteristics so that the qualitative findings are not confined to a single type of respondent. Semi-structured interviews follow a prepared guide of questions but allow the conversation to develop naturally, letting participants raise concerns the researchers might not have anticipated. The interview transcripts were then subjected to content analysis, a systematic method of coding textual data to identify recurring themes and patterns of meaning.

One of the most striking outcomes of the interview phase was the degree of convergence between participants with high anxiety and those with low anxiety. Despite sitting at opposite ends of the anxiety scale, the two groups gave strikingly similar answers in conversation. Both groups agreed that artificial intelligence cannot replace the teaching profession. That shared conviction, held across the anxiety spectrum, is a significant insight: it suggests that anxiety about AI among future science teachers is not simply a proxy for fear of professional obsolescence. Even the most anxious participants did not believe their chosen career was about to be automated away, and even the least anxious recognized that the technology raises real questions for educators.

What the interviews add, then, is a picture of anxiety that is more textured than a single scale score can convey. Moderate anxiety coexisting with confidence in the irreplaceability of teachers implies that the concerns of these pre-service educators likely center on practical and pedagogical uncertainties—how to use AI applications appropriately, what such tools mean for the integrity of learning, and how their own preparation has left them under-equipped to judge. The study’s design cannot assign precise weights to those individual concerns, but the pattern of a shared belief in the durability of the teaching profession, held by both high- and low-anxiety participants, anchors the interpretation of the quantitative result in the participants’ own words.

Based on these findings, the research team put forward two recommendations. The first is to provide pre-service teachers with the necessary training on artificial intelligence, so that the moderate anxiety observed in the survey can be addressed directly through competence-building rather than left to resolve itself. The second is to offer an artificial intelligence course to all pre-service teachers, embedding AI literacy into teacher education as a standard component rather than an elective add-on. Taken together, the recommendations treat AI anxiety not as an individual failing of nervous students but as a structural gap in teacher preparation—one that faculties of education can close by making the technology a routine, examined part of the curriculum.

The study was conducted with formal ethical oversight. All protocols and methods involving human participants were reviewed and approved by the Erciyes University Social and Human Sciences Ethics Committee under approval ID 596, and the research was carried out in accordance with relevant institutional and ethical guidelines, including the principles of voluntary participation and confidentiality. Informed consent was obtained from all participants before data collection began, and participant consent forms were required for the collection of participants’ opinions. These safeguards are standard for research involving human subjects, but they are worth noting because they underline that the anxiety being measured is self-reported, personal, and offered voluntarily by students describing their own professional futures.

For the wider conversation about artificial intelligence in schools, the study offers a timely data point. Debates about AI in education often swing between utopian predictions of personalized learning at scale and dystopian warnings about deskilled classrooms, and both extremes tend to be articulated by adults already established in their careers. This study instead listens to the people who will actually inherit the AI-infused classroom: students now training to teach science. Their measured verdict—moderate anxiety, coupled with a firm belief that no algorithm can replace a teacher—suggests that the most productive path forward is neither alarm nor complacency, but deliberate preparation. If the authors’ recommendations are heeded, the next generation of science teachers may enter the profession not anxious about artificial intelligence, but equipped to use it, question it, and keep the human core of teaching intact.

Subject of Research: Anxiety levels of pre-service science teachers regarding artificial intelligence applications in education

Article Title: Pre-service science teachers’ anxiety levels about artificial intelligence applications

Article References: Çakir, C., Gökçe, H., & Kariper, İ. A. (2026). Pre-service science teachers’ anxiety levels about artificial intelligence applications. Discover Psychology, 6(1), Article 269. https://doi.org/10.1007/s44202-026-00909-y

Image Credits: AI Generated

DOI: 10.1007/s44202-026-00909-y

Keywords: artificial intelligence, pre-service teachers, science education, teacher anxiety, mixed methods, sequential explanatory design, survey research, semi-structured interviews, content analysis, teacher training, educational psychology, Erciyes University