3. METHOD
The research was conducted at SD Negeri 1 Sinunukan, Sinunukan District, Mandailing Natal Regency, North Sumatra, Indonesia, during the odd semester of the 2025/2026 academic year, involving 250 elementary school students. This research on developing an AI literacy instrument for elementary school students is a type of study using Research and Development (R&D). Research and Development is a systematic set of steps undertaken to produce a new product or refine an existing product so that it can be accounted for by the developing researcher (Jihan, Reffiane, and Arisyanto 2019). The development of the AI literacy validation instrument in this study employed the ADDIE procedure, which includes five phases: Analyze, Design, Development, Implementation, and Evaluation (Krismony, Parmiti, and Japa 2020). The ADDIE model was chosen because it offers a systematic, structured development flow that is easily applicable in educational instrument development. Furthermore, the interconnected stages of the ADDIE model allow researchers to conduct in-depth needs analysis, design instruments tailored to the characteristics of elementary school students, develop instruments gradually, implement them in a limited setting, and evaluate them to ensure the validity, practicality, and effectiveness of the resulting instrument. Thus, the use of the ADDIE model is considered appropriate for developing an AI literacy instrument suitable for elementary school students. (Torang Siregar, & Yuni Rhamayanti. (2025)
In the Analyze phase, the researcher conducted a needs analysis regarding the aspects of AI literacy, which include the ability to recognize AI, understand AI, use and apply AI, create and evaluate AI
utilization, and exhibit attitudes and ethics in using AI. This analysis was performed to identify the students' initial conditions, learning needs, and the gap between expected competencies and existing field conditions. The problems identified in this phase served as a reference or basis for developing a self-assessment instrument as a relevant solution suited to the characteristics of elementary school students (Firda 2023). The results of this analysis then became the basis for formulating indicators, creating instrument grids, and determining the form of statement items to be developed in subsequent stages.
The Design phase aimed to design the structure of the artificial intelligence literacy instrument for elementary school students. Activities in this phase included: (1) formulating the objectives of AI literacy assessment, (2) developing indicators and sub-indicators of AI literacy covering the abilities to recognize AI, understand AI, use and apply AI, create and evaluate AI utilization, and exhibit attitudes and ethics in using AI, (3) creating an instrument grid, (4) determining the form of instrument statements, and (5) designing the scoring guidelines. The instrument design considered the alignment between indicators, statements, and the developmental characteristics of elementary school students.
The Development phase was used to compile the instrument's statement items based on the designed blueprint. Activities in this phase included: (1) drafting the AI literacy instrument, (2) conducting expert validation covering content, construct, and language validity, (3) revising the instrument based on input from experts, and (4) pilot testing the instrument with elementary school students. Data from the pilot test were then analyzed to determine expert validity, empirical item validity, expert reliability, and the reliability of the developed instrument.
The Implementation phase was carried out by applying the AI literacy instrument, which had been validated by experts and revised based on their feedback, to students in grades III and IV of elementary school. The application of this instrument aimed to determine: (1) the instrument's usability in actual learning conditions, (2) the ease with which students understood the instructions and statements, (3) the suitability of the statement items to the developmental characteristics of elementary school students, and (4) the time students needed to complete the instrument. The results of this implementation phase were used as a basis for final refinements to the instrument to make it more practical, easy to use, and aligned with the goals of AI literacy measurement.
The Evaluation phase was used to assess the final quality of the developed AI literacy instrument. This evaluation covered several aspects, namely: (1) overall construct validity, (2) instrument reliability, (3) the instrument's effectiveness in measuring students' AI literacy abilities, and (4) the feasibility of using the instrument in elementary school learning. The results of this evaluation became the basis for concluding that the developed instrument is feasible to use and can be utilized as a valid and reliable tool for measuring AI literacy in elementary school students.
The instrument developed in this study is an AI literacy assessment tool covering several aspects: the ability to recognize AI, understand AI, use and apply AI, create and evaluate AI utilization, and demonstrate attitudes and ethics in using AI. The instrument was constructed using a Likert scale to measure students' attitudes, perceptions, and levels of understanding in a graded manner. The Likert scale is an ordinal scale that allows respondents to be ranked based on their level of agreement with a given statement, although it does not indicate equal distances between responses. Each statement item has four answer alternatives: Strongly Agree (SA), Agree (A), Disagree (D), and Strongly Disagree (SD), arranged from the most positive to the most negative response (Mawardi 2019).