AI in STEAM Education: Exploring New Pathways for Student-Led Climate Action Projects

The findings suggest that students see AI as useful for creativity, independent learning and solution design, with accuracy, academic honesty and human judgment remaining central concerns.

A student can care deeply about climate change and still feel lost when asked to design a solution, especially when the task involves weather records, unfamiliar calculations and decisions about what would work in their community. A research paper titled "Empowering student-led climate solutions: the role of AI in STEAM education," published in Frontiers in Education, explores how artificial intelligence could help students navigate that gap between concern and action.

Shahinaz Osman and her co-authors examined students' experiences with AI, gathered perspectives from educators and regional experts, and proposed a teaching framework that connects climate problems with practical projects. The findings suggest that students see AI as useful for creativity, independent learning and solution design, with accuracy, academic honesty and human judgment remaining central concerns.

Students have the tools, but climate action needs more support

The research combined a review of 80 studies with a survey of 69 undergraduates enrolled in an AI education course at United Arab Emirates University. All 82 eligible students were invited, producing an 84.1% response rate. Interviews brought in 25 participants: 11 policy and sustainability experts, nine faculty members and five student innovators.

Every surveyed student reported using AI in education, and 75.4% used it daily or weekly. Formal AI training had reached 71% of respondents, and 82.6% believed AI should be integrated into education. Climate concern was universal among the surveyed students, but participation in practical activities was much less common. Only seven students, about 10.1%, reported involvement in the university's climate project, and 14.5% had participated in student-led projects or hackathons.

The gap matters because access to a chatbot does not automatically give students a project to lead, reliable environmental data or a route to share their ideas with decision-makers. The researchers describe a need for learning experiences that give students responsibility for choosing problems, investigating evidence and developing responses.

Students' answers revealed where they felt AI helped most. Creativity and independence received the strongest agreement, at 86.9%. Feeling empowered to propose climate solutions followed at 76.8%, and 73.9% said AI helped them move from collecting data toward strategic solution design. Perceived support for critical thinking reached 68.1%; better academic work received 65.2%; more personalised and accessible learning received 59.4%.

Giving students room to think beyond the data

Climate projects can become overwhelming when students must interpret large datasets before they can explore an idea. The authors describe AI as a "Complexity Mediator," a tool that organises information, supports experimentation and handles some computational work so students have more attention available for meaningful decisions.

Their explanation draws on distributed cognition, the idea that thinking happens through interactions among people, tools and information. In a classroom, that could mean students using AI to explore patterns in environmental records, discussing those patterns with classmates and checking their interpretation with a teacher.

The paper's review identified four main types of tools:

• conversational chatbots for questions and brainstorming,
• generative AI for visualising scenarios,
• machine learning for analysing data, and
• systems that connect AI with physical sensors for resource management.

Different tasks require different tools; generating a persuasive explanation and predicting water consumption involve different kinds of evidence. STEAM brings science, technology, engineering, arts and mathematics into the same learning experience. The arts contribute creative design, ethical reflection and communication, helping students explain why a solution matters to the people expected to use it.

Interviewed students described AI as a technical partner that made difficult prototyping feel more accessible. Faculty members emphasised the importance of preserving thoughtful inquiry, since a fast answer can easily be mistaken for an accurate one.

A four-step route from local problems to practical proposals

The researchers propose the AI-Empowered STEAM Climate Action Model, or AISC-AM, to give student projects a clear structure.

Observe: Students identify a local environmental problem and gather evidence. Image recognition or sensor tools could help monitor pollution, identify invasive species or investigate air quality, giving the project a foundation beyond personal impressions.

Create: Learners develop a physical or digital prototype using tools suited to their problem. A smart irrigation system could combine sensors, weather information and machine learning to explore ways of reducing water waste.

Validate: Students test assumptions and estimate possible impacts, including water savings, energy reductions or the feasibility of expanding a prototype. Predictions need checking against regional environmental data and scientific principles.

Communicate: Teams examine bias and prepare accessible explanations, policy briefs or public awareness materials that make their evidence understandable to a wider audience.

The paper describes project outputs involving campus water optimisation, urban heat mapping and regional climate advocacy. Its proposed assessment dimensions cover scientific accuracy and local relevance, technical innovation, feasibility and scalability, and ethical validation.

Illustrative examples include pollution dashboards and school energy proposals. Their predicted benefits should be read as possibilities to investigate, rather than verified environmental savings. The framework offers a proposed teaching pathway supported by preliminary stakeholder feedback.

Human checks and real partnerships make the difference

Students liked using AI but recognised its risks; Accuracy and reliability concerned 60.9% of respondents, academic honesty and plagiarism concerned 59.4%, privacy and data protection concerned 52.2%, and overdependence concerned 43.5%.

The authors recommend teaching students to check sources, spot bias, and compare AI answers with real evidence. Their "AI-in-the-Loop" approach puts students and teachers in charge of these checks. One student checked AI advice against local climate reports because it might misunderstand UAE soil conditions.

Experts identified funding shortages, limited institutional support, political tensions and weak links between university research and policymaking. Partnerships with organisations such as the Arab Youth Center could build students' green skills and help their proposals reach decision-makers.

The study covered one university and one course, with many participants studying education. It measured students' opinions without comparison groups or before-and-after skill tests, so it cannot prove that AI improved skills or that the findings apply elsewhere.

The researchers recommend longer studies that compare groups, test climate knowledge and ethical judgment, and assess teacher training. Practical success depends on students checking their ideas carefully and developing solutions communities can use.