Learning algebraic concepts, particularly algebraic identities, is one of the major challenges in mathematics education at the lower secondary level. This study aimed to investigate the effectiveness of computer-supported collaborative learning with the aid of AI-based group chats on ninth-grade students’ understanding of algebraic identities, based on the APOS theoretical framework. The research employed a mixed‑methods (quantitative–qualitative) design. In the quantitative phase, a quasi-experimental pretest–posttest design with a control group was used. The sample consisted of 60 ninth-grade male students from Kerman city in the 2024–2025 academic year, selected through cluster random sampling and assigned to experimental and control groups. The experimental group received instruction over two weeks in four consecutive sessions using computer-supported collaborative learning with AI-assisted group chats, while the control group received traditional instruction. The research instrument was a researcher-made test based on APOS levels. The findings indicated a significant improvement in the experimental group’s mean scores, with a statistically significant advantage over the control group and a large effect size (p < 0.001). The qualitative analysis of interactions also revealed cognitive development from the Action level toward the Process, Object, and Schema levels. Overall, integrating artificial intelligence into collaborative learning contributed to deepening students’ conceptual understanding of algebraic identities.