EMPOWERING ELEMENTARY SCHOOL TEACHERS THROUGH MATHEMATICAL PROBLEM POSING WITH AI AS A REFLECTIVE FEEDBACK TOOL: A COMMUNITY SERVICE PROGRAM IN DEPOK CITY

Program pengabdian kepada masyarakat ini menangani tantangan dalam desain instruksional di kalangan guru sekolah dasar di Kota Depok, di mana kemampuan untuk menyusun soal aritmetika yang bersifat komputasional belum mencakup perancangan tugas yang secara matematis autentik dan menuntut kemampuan kognitif tinggi. Inisiatif ini bertujuan untuk mendampingi 18 guru sekolah dasar dalam merancang, mengevaluasi, dan menyempurnakan tugas-tugas matematika melalui siklus penyusunan soal secara partisipatif, dengan memanf…

By the Truvace deskOctober 20, 20261 min read1 sourcePeer-reviewedTRV-2026-2001· cite & verify

In brief

A community service program in Depok City worked with 18 elementary school teachers who could write computational arithmetic problems but did not routinely design mathematically authentic, cognitively demanding tasks. Through a participatory problem-posing cycle, teachers used AI as a reflective feedback partner to design, evaluate and refine tasks.

The qualitative evaluation points to a shift in the quality of the tasks teachers produced, which suggests AI feedback can support instructional design training. The authors stress that this is not verified evidence of improved teacher competence, and effects in actual classrooms remain untested.

Main points

The program involved 18 elementary school teachers in Depok City.
The evaluation examined four initial group problems, collaborative group revisions and nine fully documented individual products.
Baseline tasks were routine single-step computations with superficial narrative wrappers or no context.
Refined tasks more often involved multi-step, analytical or evaluative reasoning in realistic-generic or local contexts.

The gain

After AI-assisted collaborative appraisal and individual refinement, most documented teacher-written math tasks shifted from routine computation to multi-step or analytical reasoning in realistic or local contexts.

The rundown

The initiative targeted a specific gap in instructional design: teachers could construct computable arithmetic items, but this ability did not routinely extend to tasks that were mathematically authentic and cognitively demanding.

Teachers followed a participatory problem-posing cycle in which AI acted as a reflective feedback partner during collaborative appraisal, followed by individual refinement of each task.

The evaluation was qualitative and descriptive, comparing baseline group artifacts with revised group and individual products rather than measuring teacher skills directly.