Artificial Intelligence in Science Learning within the Framework of Situated Learning Theory: A Qualitative Investigation of Teachers' Perspectives

Synthesis: Shang Li (2025) presents a qualitative study grounding AI integration in K-12 science education within Situated Learning Theory (SLT). Drawing on semi-structured interviews and open-ended questionnaires with fourteen Chinese science teachers (elementary through high school; physics, mathematics, chemistry, and general science), the study investigates how AI tools influence students' learning processes and outcomes, transform classroom interaction and social construction, and reshape teachers' pedagogical identities. Findings show that virtual labs, simulations, and intelligent tutoring systems enhance conceptual understanding and higher-order scientific skills, cultivate scientific identity through contextualized low-risk inquiry, and deepen collaborative meaning-making. Teachers' roles shift from knowledge transmitters to facilitators, co-investigators, and ethical supervisors, while teachers simultaneously voice concerns about overreliance and the erosion of core scientific abilities. The study positions AI tools as "mediational artifacts" that extend SLT by enabling digital communities of practice and boundary-crossing between school, real-world, and interdisciplinary contexts.

Key Findings

• AI enhances conceptual understanding and higher-order scientific skills. Dynamic simulations and visualizations lower cognitive barriers and deepen comprehension beyond rote memorization, while supporting experimental design, inquiry, hypothesis testing, data analysis, modeling, argumentation, and creative exploration.
• AI transforms learner identity and Learner Agency. Students transition from "knowledge learners" to "scientific practitioners," internalizing autonomy, confidence, and a scientific identity through authentic tasks and low-risk experimentation in Feedback-rich environments.
• AI deepens classroom interaction and social construction. AI-enabled real-time feedback and shared data foster collaborative inquiry, role division, and critical conversation ("thinking-negotiating-optimizing"), while teachers move to facilitators, questioners, and co-investigators and take on new roles as ethical supervisors of responsible AI use.
• AI enables authentic, contextualized, and boundary-crossing learning. Virtual labs and simulations let students experience distant, hazardous, or imperceptible phenomena, bridge virtual and real-world practice, and connect scientific knowledge to real-life and cross-disciplinary contexts.
• Teachers are optimistic yet concerned. Teachers envision immersive, personalized AI futures but worry about being replaced and about overreliance degrading independent thinking, hands-on experimentation, and Problem Solving abilities.

What this means for practice

• Instructors. Score students' unaided reasoning alongside the AI-assisted artifact, because the interviewed teachers paired praise for simulations with fear that overreliance erodes hands-on experimentation and Problem Solving.
• Instructors. Assign rotating roles across an AI-supported inquiry — hypothesis designer, data analyst, and critical questioner of the model's output — so AI works as a mediational artifact rather than as the source of answers.
• Instructors. Run a short audit task in which students check an AI-generated scientific claim for sourcing, bias, and data provenance, mirroring the "ethical supervisor" role the teachers described taking on.
• Teacher educators. Have candidates design a lesson whose stated teacher role is co-investigator rather than transmitter, then debrief how it changed the questions they asked, since the fourteen teachers framed this identity shift as their central adaptation to AI.
• Teacher educators. Rehearse how to justify and sequence AI use within a unit rather than defaulting to it, addressing the replacement anxiety and authenticity concerns the teachers raised.

Limitations

• Fourteen teachers were interviewed, recruited through the Chinese social media platform Red, whose user base skews toward younger adults; the authors flag a possible generational bias because these teachers were more open to technology than the wider STEM teaching population.
• Participants spanned elementary through high school, and the authors note that conceptual development, curricular focus, and learning goals likely differed substantially across those levels, limiting comparability.
• The study did not record whether students engaged with AI individually, collaboratively, or as a whole class, nor the frequency or duration of use, so the authors cannot rule out that disparities in technological mastery shaped the reported outcomes.
• Teachers drew on different tool types — generative AI, virtual labs and simulations, or both — which the authors state limited the comparability of their experiences and interpretations.

Citation

Li, S. (2025). Artificial Intelligence in Science Learning within the Framework of Situated Learning Theory: A Qualitative Investigation of Teachers' Perspectives. Creative Education, 16(11), 1858–1882.