An Inductive Typology of Artificial Intelligence Use in Science Teaching and Its Alignment with AI Literacy Frameworks

Artificial intelligence has moved from the realm of science fiction into the daily conversation of schools, promising to reshape how children learn and how teachers work. In the specific world of science education, where students learn by asking questions, building models, and testing ideas, the arrival of these smart computer systems raises a fundamental question: how are they actually being used? The conversation often swings between two extremes. On one side, there is the hope that these tools will revolutionize learning, acting as personal tutors for every student. On the other, there is the worry that they will simply automate tasks or replace human judgment. To understand what is really happening, researchers must look past the hype and examine the actual practices described in academic studies. They need to know if teachers are using these tools to create new ways of thinking about science, or if they are simply fitting them into old routines. This is the territory explored by a recent large-scale study that sought to map the real-world landscape of artificial intelligence in science classrooms.

To make sense of this complex field, the researchers began with a massive collection of evidence. They gathered 795 peer-reviewed journal articles that discussed artificial intelligence in the context of science teaching. Instead of reading every single word of every paper, they focused on the abstracts—the short summaries that describe the main point and methods of each study. This allowed them to see the broad patterns across thousands of pages of research without getting lost in the details of any single experiment. Their goal was not to test a specific tool or to prove that one method works better than another. Instead, they wanted to build a picture of how teachers are currently described as using these technologies. They approached the data without a pre-made list of categories, letting the patterns emerge naturally from the text itself. This method ensured that their findings reflected what was actually being reported in the literature, rather than what researchers expected to find.

As they sorted through the hundreds of summaries, six distinct ways of using artificial intelligence began to stand out. The most common use, appearing in nearly 78 percent of the studies, was for instructional delivery and pedagogy. In these cases, teachers used artificial intelligence to help present content, guide students through inquiry, or explain difficult concepts during class. The second most frequent category involved assessment and feedback, found in about 46 percent of the papers. Here, the technology helped teachers check student work, grade assignments, or provide immediate comments on performance. The third group, appearing in roughly 24 percent of the studies, focused on lesson planning and curriculum design, where teachers used the tools to prepare materials or organize their courses before the students even arrived. The remaining categories were less frequent but still significant: using interactive agents like chatbots to talk with students, employing simulations and virtual environments to model scientific phenomena, and utilizing the technology for the teachers' own professional development and support.

The researchers then took these six categories and looked at them through a specific lens: a framework developed by the Organisation for Economic Co-operation and Development to describe how people engage with artificial intelligence. This framework divides engagement into four areas: interacting with the system, creating content with it, managing its outputs, and designing the system itself. When the researchers mapped their findings onto this structure, a clear story emerged. The literature showed that teachers are overwhelmingly using artificial intelligence to interact with it and to design how it fits into their lessons. They are not typically building the underlying computer code or the complex algorithms that power these systems. Instead, they are acting as skilled integrators who decide how to apply these tools to their specific teaching goals. They configure the tools, adapt them to their classroom needs, and structure the learning activities around them.

This distinction is crucial because it challenges the idea that teachers need to become software engineers to use artificial intelligence effectively. The study suggests that the primary role of the teacher in this new landscape is that of a pedagogical architect. They are the ones who interpret the outputs of the technology, decide when to use it, and ensure it aligns with the goals of science education, such as fostering evidence-based reasoning or understanding complex models. The research indicates that while the technology is powerful, its educational value comes from how teachers weave it into their existing practices. The studies rarely described teachers as passive recipients of technology, nor did they show them as the technical creators of the systems. Instead, they were portrayed as active decision-makers who use these tools to enhance their professional judgment.

The findings also revealed that the use of artificial intelligence in science education is largely conservative in the best sense of the word. Rather than overturning the fundamental structures of how science is taught, the technology is being absorbed into established routines. It is used to support the core tasks of teaching: explaining concepts, checking understanding, and planning lessons. While there is excitement about the potential for radical change, the current reality is one of augmentation. Teachers are using these tools to handle the heavy lifting of assessment or to generate materials for planning, freeing up time to focus on the human elements of teaching. The research suggests that the most common use of artificial intelligence is to make the existing work of a science teacher more efficient and effective, rather than to replace the teacher or to create an entirely new way of learning.

One of the most interesting aspects of the study is what it says about the future of teacher training. Because the literature shows that teachers are primarily using these tools to design and manage learning experiences, the study implies that professional development should focus on these skills. Teachers need to learn how to critically evaluate the outputs of artificial intelligence, how to adapt these tools to their specific subject matter, and how to integrate them responsibly into their curriculum. They do not necessarily need to learn how to code the systems themselves. The study highlights that the most valuable skill for a science teacher in the age of artificial intelligence is the ability to judge when and how to use these tools to support the unique demands of scientific inquiry.

The researchers were careful to note the limits of their work. Their findings are based on what is written in academic papers, which may not perfectly match what happens in every classroom. They also noted that their analysis focused on English-language publications, which might miss some local or non-English perspectives. Furthermore, because they analyzed summaries rather than full texts, they captured the broad strokes of how these tools are described, but not the deep, nuanced details of every single lesson. Despite these limitations, the study provides a solid, evidence-based map of the current situation. It moves the conversation away from speculation and toward a clear understanding of how artificial intelligence is actually being positioned in science education today.

In the end, the study paints a picture of a profession that is adapting to a new tool with practical wisdom. Artificial intelligence is not arriving as a mysterious force that will dictate the future of education. It is arriving as a resource that teachers are learning to use, shape, and direct. The research shows that teachers are not waiting for technology to solve their problems; they are actively figuring out how to use it to solve the problems they already face. By focusing on the actual practices described in hundreds of studies, the researchers have shown that the integration of artificial intelligence in science teaching is a story of human agency. It is a story of teachers using new tools to do their jobs better, ensuring that the technology serves the goals of learning rather than the other way around.