MODEL PEMBELAJARAN DEEP LEARNING DAN HUMAN-AI COLLABORATION DI KELAS SEKOLAH DASAR: TINJAUAN SISTEMATIS

Abstract

Deep learning and human-AI collaboration models have become a policy direction in Indonesian
primary education following the adoption of meaningful learning approaches within the Kurikulum
Merdeka in 2025-2026, in line with the global trend of integrating artificial intelligence (AI) into
primary curricula. However, systematic studies examining the characteristics, effectiveness, and
implementation challenges of these two integrated approaches at the elementary school level remain
limited. This study aims to analyze and synthesize prior research findings regarding the design,
implementation, and impact of deep learning models supported by human-AI collaboration in
elementary classrooms, and to map research gaps relevant to the Indonesian context. This study

employed a systematic literature review method following the PRISMA protocol, searching Scopus,
ERIC, Google Scholar, Garuda, and SINTA databases using the keywords deep learning elementary
education, human-AI collaboration classroom, and AI literacy primary school, covering publications
from 2020 to 2026. A total of 19 articles meeting the inclusion criteria were thematically analyzed and
coded into four theme categories. The coding results show that the characteristics of deep learning models
dominated the findings (36.8%), followed by forms of teacher-AI collaboration (31.6%), impact on
learning outcomes and motivation (21.1%), and challenges and supporting factors for implementation
(10.5%). Project-based learning and group collaboration supported by adaptive AI tools are most
consistently associated with improved conceptual understanding, motivation, and digital literacy
among elementary students, although success depends heavily on teacher readiness, infrastructure
availability, and alignment with children's cognitive development stages. Key challenges include
limited teacher training, digital infrastructure gaps, and the absence of clear ethical guidelines for AI
use among elementary-aged students. This study recommends the development of co-design-based
teacher training models and contextual ethical frameworks as a foundation for AI integration policy in
Indonesia's elementary curriculum