8

Instructional Approaches and Professional Practice

Computational thinking, data science, and artificial intelligence (AI) can fundamentally shift instructional practices by adapting, augmenting, or supplementing traditional modes of teacher and student work in classrooms. Several earlier presenters, for example, described how generative AI can assist teachers in developing lessons and assessments, while becoming a tutor or partner for students in solving more complex real-world problems.

Furthermore, teacher preparation and sustained professional learning remain the most decisive factors in translating ambitious policy into classroom practice. In many countries, these efforts are treated not as optional support but as essential policy infrastructure that underpins successful implementation. The most effective initiatives embed professional learning into curriculum development cycles, allowing teachers to pilot materials, provide feedback, and build confidence with emerging content and tools. This tight linkage between curriculum design and capacity building increases the likelihood that instructional shifts are durable and scalable.

In the final session, moderated by Hollylynne Lee, presenters from Singapore, Germany, Japan, Finland, and Turkey discussed changes in professional development and teaching practice in the classroom. The following questions were addressed by the panelists:

1. What are the theoretical perspectives and contemporary research by professional scientists on AI, computational thinking, and data science that have implications for the use of these tools for educational purposes?
2. What are the current practices and findings on the teaching of AI, computational thinking, and data science with K-12 students across the world?

3. What knowledge and teaching practices should educators develop about how to integrate AI, computational thinking, and data science within mathematics learning in K-12?

SINGAPORE: PRACTICES IN COMPUTATIONAL THINKING, DATA SCIENCE, AND AI

Tin Lam Toh (National Institute of Education, Nanyang Technological University) noted that teachers in Singapore may not see computational thinking, data science, and AI explicitly in the curriculum documents. However, evidence of student exposure to these is present, starting with “Code for Fun” and “AI for Fun” enrichment activities, both of which expose students to basic concepts and tools. Secondary students are exposed to computational thinking and programming skills prior to their O-levels, the high-stakes national examination taken in grade 9.

He considered AI in two aspects of learning and teaching. An Adaptive Learning System was adapted for student learning by personalizing learning through customizing an appropriate learning trajectory for each student. AI-enabled digitized textbooks are being developed with possibly an AI-powered feedback system. Although teachers are spared repetitive grading through AI, they remain important to ensure students’ learning and are expected to review all AI-generated feedback provided to students. Even before the AI era, Singapore schools had been exploring other means of adaptive feedback. In professional development, teachers have drawn on their pedagogical content knowledge to develop game-based and other types of assessment.

Moving on to AI for teachers, Toh described that AI-generated lesson plans could be used to assist teachers, rather than teachers having to start from scratch. It can remove some of the mundane work. Teachers provide input, and then critique and adjust the output. The focus can be on higher-order thinking skills. It responds to the Ministry of Education’s AI in education ethics framework (presented by Cheong Kang Hao in Chapter 3).

Computational skills—abstraction, decomposition, algorithmic thinking, and generalization—are “parked” under mathematics, Toh said. Spreadsheets are a common feature, with coding introduced for some students. Teachers are expected to know how to use Microsoft Excel. He also pointed to close collaboration between classroom teachers and mathematics educators. While data science is not an independent subject, the mathematics curriculum has increased the emphasis on data collection, representation, and interpretation (not just calculations), and there will likely be a revamp of the secondary school statistics strand within the school mathematics curriculum. Finally, he mentioned a learning resource called the Student Learning Space. It is a one-stop platform in response to

Singapore’s EdTech Masterplan 2030 and it is where all AI tools can be found, among many other resources for teaching and learning.

These changes have modified teacher education and professional development. A bachelor’s degree in mathematics with direct teaching qualifications is now renamed as mathematics and computational thinking, to reflect the new skills required for mathematics teachers. In mathematics teacher education offered by the National Institute of Education (Nanyang Technological University), the statistics component has gained importance. All mathematics pre-service teachers from the bachelor program with direct teaching qualification learn statistics starting from their first year, and data science is now one of the core courses for a master of science (mathematics for educators) degree offered by the National Institute of Education.

GERMANY: THE PRODABI PROJECT

Rolf Biehler (Paderborn University) related ideas about mathematics education that have been developed through the ProDaBi Project, which supports teaching of data science and big data at the school level.

As noted by Steffen Schneider earlier (see Chapter 7), Germany’s 16 states develop their own curricula based on national standards. Probability and statistics are in mathematics education across states, although to varying degrees. Computer science is becoming obligatory in a growing number of state curricula. AI, but not data science as such, is becoming part of most curricula on computer science education. Elective courses on data science are being developed for grades 9, 10, and 12, however, that are not constrained by curricular requirements.

An active community of data science and stochastics educators has formed. He and a colleague launched a working group of representatives from different educators and researchers from school subjects under the heading of data literacy. Fourteen associations have expressed interest. In addition to the innovative projects described in other sessions (see Chapters 3, 4, 5, and 7), Project QuaMath is a large effort for all topics and levels related to improving teaching practices rather than generating new content.¹ Several modules relate to teaching probability and statistics. QuaMath has involved more than 8,000 teachers and 2,000 schools as of January 2026.

ProDaBi is a collaboration between computer science education and mathematics education. Instructional strategies engage students in authentic data practice; connect data, modeling, and algorithmic systems conceptually; support generalizing transfer to other contexts; and make ethical and political dimensions an explicit part of education. Teaching units have mainly been developed for what are called elective compulsory courses

___________________

¹ For more information, see https://www.leibniz-ipn.de/en/research/projects/quamath

(students can choose specific courses from a mandated list); some, he noted, are available in English.²

Biehler reflected on the challenges faced, all of which have become part of ProDaBi’s professional development courses for teachers. First is what he called “triple content expansion,” because the topics that students need to learn involve mathematics, computer science, and other domain knowledge. In addition, deeper content knowledge is needed for transfer and generalization, which requires more teacher understanding. There are tensions with traditional curricula and teachers’ mindsets, which must be dealt with. In addition, mathematics and computer science educators are not used to moderating ethical and political discourse in the classrooms. He presented several examples of the tools and topics related to data science and machine learning, in particular a unit involving data on climate change.

FINLAND: COMPUTATIONAL THINKING, DATA SCIENCE, AND AI

Markku Hannula (University of Eastern Finland) began with what he called a common misunderstanding. “We think we know what we think and understand,” he commented. But a lot of below-the-level processing is going on based on individual experiences. He said this is a starting point for him. For that reason, not all aspects of learning are really known. His own perspective on learning is both cognitive and social. This is important when considering early math learning, for example, in which students clap or stomp in patterns. People learn with and through their bodies, which places some limitations on how much AI can support learning, Hannula said. It is important to have multiple representations or modalities to learn, which AI can help provide. The social aspect is important, learning from and with other people. Being part of the community is one of the aims of learning.

As others have said, computational thinking, data science, and AI are rapidly developing on a fast-moving train. Any knowledge will be outdated soon, which means that process competencies are increasingly important and learning to learn is essential. In addition to content knowledge, the Finnish national mathematics curriculum stresses a positive disposition, problem-solving and thinking competencies, and learning to learn. Specific technology receives little attention and time in primary education, although more in upper grades. However, the requirements of the examination for students in grade 12, which is based on a digital assessment platform, means that tasks related to some technological tools are moving down to

___________________

² See https://www.prodabi.de/en/

other grade levels. He provided an advanced mathematics problem that was developed for the 2021 examination.

Outlining the potential and risks of AI in education, Hannula concurred that AI can serve as a teacher’s assistant as well as an assistant teacher for students and parents. It is patient, present all the time, and adaptive to students’ needs. Research shows students’ attitudes and motivations improve with introduction of an AI tool. But he cautioned against students relying on AI as their answering machine or teachers letting AI take charge. AI cannot replace human contact.

JAPAN: ROLE OF MATHEMATICAL MODELING AND CLASSROOM PRACTICE IN GRADE 7

Takashi Kawakami (Utsunomiya University) said mathematical modeling is emphasized in the current Japanese mathematics curriculum and has a potential role in integrating computational thinking, data science, and AI into classroom practice. Mathematical modeling is a scientific process involving the generation, evaluation, and revision of mathematical models. As an interdisciplinary practice, it can be regarded as a boundary object of computational thinking, data science, and AI. Modeling is part of math activities at different levels.

Kawakami has been involved in “data, models, and modeling in STEAM” research, a collaborative professional development program between a math teacher and himself as math education researcher. The purpose is to provide age-appropriate experiences in an unplugged environment. It involves a grade 7 class that spends five 50-minute lessons in mathematics classrooms involved in an image classification task around the school mascot. Students create models to determine whether generated images are real or fake. The teacher emphasizes “mathematization” of daily or social phenomena, which has long been a goal of Japanese mathematics education. The task encompasses (1) data generation and data-driven activities that are part of data science, (2) algorithmic representations and reasoning that are part of computational thinking, and (3) awareness of model assumptions and limitations that help with a fundamental understanding of AI.

A challenge for Japan is aligning the core topics of computational thinking, data science, and AI with the future mathematics curriculum, which is crucial for enabling teachers to incorporate such teaching practices into their regular math lessons. Professional learning in Japan related to mathematical modeling, statistics, and data science in K-12 mathematics includes educational resources, seminars and workshops, and a certification exam on knowledge and application skills in statistics and data science.

TURKEY: CREATING SPACE FOR DATA SCIENCE IN K-12 MATHEMATICS EDUCATION

Sibel Kazak (Middle East Technical University) said beginning with the 2024–2025 academic year, the Century of Turkey Educational Model has an explicit emphasis on competencies in mathematics domains and literacy.³ Within mathematics, students do not just perform calculations. They learn mathematical reasoning, problem solving, representations, and mathematical tools and technology, as well as working with data and making data-driven decisions. The model explicitly encourages interdisciplinary connections, has a strong focus on digital literacy and technology integrations, and highlights inquiry-based and real-life problem solving. In grade 12, for example, students work with existing data to design and carry out a statistical investigation and share their results. There are, however, clear limitations. Data science is not set out as an explicit curricular strand and AI is not systematically integrated into mathematics, and computational thinking only appears indirectly.

What is needed is not new topics but new classroom practices, Kazak argued. Rather than tidy, ready-made datasets, she called for messy, unstructured data; from predefined variables to students generating their own variables; from procedural classifications to inquiry and modeling; from static representations (such as charts in textbooks) to digital data visualization tools; and from a focus on the answer to a focus on interpretation, prediction, and decision making. She described the Professional Data Science Education for Teacher Education at the University Level project to prepare teachers for this kind of instruction.⁴ It was funded by the European Union and involved six higher education institutions in five countries, including Turkey. They are developing data science education modules with the integration of digital tools into STEAM subjects. The modules are based on the Empowering Data Science Understanding for Teacher Education Conceptual Framework and aimed at pre-service teachers and their instructors.⁵

The modules provide structured scaffolding, working with “messy” data, AI and machine learning activities, embedded computational thinking,

___________________

³ Millî Eğitim Bakanlığı. 2024. The Century of Türkiye Education Model Teaching Programs Common. Ministry of National Education. https://tymm.meb.gov.tr/upload/brosur/ortak_metin.pdf

⁴ For more information on the project, see https://datasetup.euc.ac.cy/

⁵ For more information on the framework, see A. Leavy et al. 2026. Designing data science learning in initial teacher education: The EDUCATE Conceptual Framework. Education Sciences, 16(2). https://doi.org/10.3390/educsci16020307

and ethical considerations.⁶ In the Turkish context, challenges include a high-stakes assessment culture, limited formal preparation in data science and AI, low teacher confidence in guiding open-ended exploration, and limited experience with digital tools, and not enough resources for teachers.

DISCUSSION

Hollylynne Lee noted the visions across classrooms with different instructional approaches. She asked presenters to imagine exemplar experiences in their countries. Hannula suggested that teachers have to be prepared that the classroom will be noisy and nonstructured as students work on projects in groups. The teachers’ key role would help in metacognitive and affective regulation for the groups, stepping away from their traditional role of cognitive guidance. Biehler said teachers’ challenges are different depending on the topic. ProDaBi’s teacher development courses offer time in between the days of training so that teachers can implement what they learn back into their classrooms and then return to the training to report on the challenges. They may be learning the new content themselves, in addition to learning how to teach it. They also receive background materials in a usable format, for example, about climate change. Kazak highlighted the interdisciplinary aspect of teaching data science. She said it is difficult for mathematics teachers to know the context in other subject areas. Toh said computational thinking should be infused in teacher training, moving from calculations and theorems to the processes behind them. Kawakami said, in his example in Japan, teachers need to be able to learn and experience mathematical modeling processes as a priority.

Matti Tedre commented that the use of ready-made datasets and student-generated data are both used, and asked about the pedagogical angle that might speak to either of these directions. Hannula said it is very motivating when students can generate their own data. The ready-made data-set, however, reduces the time involved. Biehler suggested a middle area, between data that students gather themselves locally at a few points and a global level. For example, students from different schools can gather data so they are engaged themselves but the datasets are larger to see trends and relations. Kazak agreed there are advantages to both and they gain different skills. When they use ready-made sets, they also have to develop skills in interrogating the data.

When asked by an online attendee about longitudinal effects of using their tools and frameworks or, if not known, what needs to be studied

___________________

⁶ For more information on the modules, see M. Meletiou-Mavrotheris et al. 2025. Investigating student teacher engagement with data-driven AI and ethical reasoning in a graduate-level education course. Education Sciences, 15(9). https://doi.org/10.3390/educsci15091179

longitudinally, Biehler pointed to some meta-analyses that show that students are eager to engage in new ways of teaching. He noted some research around constructing decision trees that have been helpful. Conrad Wolfram said his organization has many ready-to-use datasets that are particularly useful for students as they are beginning to learn, recognizing that they must also learn to handle messier data. Padhu Seshaiyer commented that students want to solve societal problems. He asked about ways to connect data from the United Nations Sustainable Development Goals into a mathematics curriculum. Biehler said the climate change unit he highlighted was developed by biology and data science teachers, and has worked well as an elective compulsory course in grade 9.

Zarek Drozda said status quo math education has often fallen into the camps of procedural fluency and conceptual understanding, and the debate has occurred for decades. He asked, in working with educators and new technologies, if a third camp needs to be considered or perhaps a more holistic balance. Toh said Singapore believes they mutually reinforce each other. It is not either–or. In primary school education, procedure is emphasized at first, but then students get to the concepts and ask why. Kawakami said both are connected in the Japanese context.

David Weintrop asked about challenges for teachers to adopt and bring new ideas into the classroom and ways to overcome the challenges. Toh related that, when he and a colleague embarked on the computational thinking project described above, they realized that many teachers could not use Excel, despite the assumption that they could. Instead of coding, a lot of time was spent on basic spreadsheet functions. Biehler said there is a tendency in some classrooms to just achieve a certain task with digital tools. The next steps of interpreting and using the tools are important, and examples help. New types of assessments were invented for the climate change unit he described, such as creating social media posts (internal) on what students have learned. Teachers should be prepared to spend more time at the beginning with asking questions, which is difficult for some traditional teachers to do.

This page intentionally left blank.