<![CDATA[The massive use of generative artificial intelligence (AI) among university students is not matched by adequate AI literacy skills in validating scientific references. This condition gives rise to a serious threat in the form of academic citation hallucination, a phenomenon in which AI produces fictitious references that appear convincing but cannot be found in any scientific database. An initial survey of 30 students at the Elementary School Teacher Education Department (PGSD), Faculty of Teacher Training and Education (FKIP), Universitas Halu Oleo (UHO) in January–February 2026 revealed that 80% had used ChatGPT to compile references, 70% had never verified their validity, and 90% were unaware of verified academic AI tools. A review of 20 academic papers found that 35% contained references untraceable in any scientific database. This community service activity aimed to improve students' AI literacy in validating scientific references through online training using three academic AI tools: Scite AI, Elicit AI, and Consensus AI. The method employed was participatory training based on live demonstration and guided practice, using a one-group pre-test post-test design (n=30), analyzed using the Hake N-Gain formula. Results showed an average score increase from 51.40 (pre-test) to 87.33 (post-test) with a mean N-Gain of 0.73 (High category). N-Gain distribution: 15 participants (50%) High and 15 participants (50%) Medium, with no participants in the Low category. Participant satisfaction reached 83.33% (Very Satisfied). Online academic AI tools training is proven effective in improving AI literacy among prospective elementary school teacher students in validating scientific references.]]>
Copyrights © 2026
