<![CDATA[Abstract: Post-pandemic literacy-numeracy learning losses remain a challenge for primary education in Indonesia, while evidence of the use of Artificial Intelligence (AI) as a recovery strategy has not been specifically synthesized. This study aims to synthesize empirical evidence regarding the use and effectiveness of AI in supporting the recovery of primary school students' literacy-numeracy and realizing its empowerment for Indonesia. The study used a Systematic Literature Review (SLR) with the PRISMA protocol. From 264 records obtained through Scopus, Elicit, SciSpace, and Google Scholar, 17 studies met the inclusion criteria, consisting of 12 primary empirical studies and 5 systematic observations/meta-analyses spanning 2019–2026. The results indicate that AI has the potential to support recovery through personalization, adaptive feedback, and learning support tailored to student needs. Some studies reported effect sizes of 0.60–0.73, but these figures do not represent the full range of research. Its implementation affects teacher readiness, infrastructure, equitable access, and ethics and data protection. These findings form the basis for recommendations for contextual AI implementation for primary education in Indonesia.Abstrak: Learning loss literasi-numerasi pascapandemi masih menjadi tantangan pendidikan dasar di Indonesia, sementara bukti pemanfaatan Artificial Intelligence (AI) sebagai strategi pemulihannya belum tersintesis secara khusus. Penelitian ini bertujuan mensintesis bukti empiris mengenai pemanfaatan dan efektivitas AI dalam mendukung pemulihan literasi-numerasi siswa sekolah dasar serta merumuskan implikasinya bagi Indonesia. Penelitian menggunakan Systematic Literature Review (SLR) dengan protokol PRISMA. Dari 264 record yang diperoleh melalui Scopus, Elicit, SciSpace, dan Google Scholar, diperoleh 17 studi yang memenuhi kriteria inklusi, terdiri atas 12 studi empiris primer dan 5 tinjauan sistematis/meta-analisis pada rentang 2019–2026. Hasil menunjukkan AI berpotensi mendukung pemulihan melalui personalisasi, umpan balik adaptif, dan dukungan belajar sesuai kebutuhan siswa. Beberapa studi melaporkan effect size 0,60–0,73, tetapi angka tersebut tidak merepresentasikan keseluruhan studi. Implementasi dipengaruhi kesiapan guru, infrastruktur, pemerataan akses, serta etika dan perlindungan data. Temuan ini menjadi dasar rekomendasi implementasi AI yang kontekstual bagi pendidikan dasar Indonesia.]]>
