User Acceptance and Effectiveness of AutoML Systems in Predicting Training Success: A Case Study of the DTS Program

Abstrak

The advancement of digital technologies has transformed workforce demands, emphasizing digital literacy, adaptability, and innovation. Indonesia’s Digital Talent Scholarship (DTS) program, launched in 2018, has trained over 500,000 participants in IT skills. However, completion rates vary significantly (57%–96%), highlighting challenges in participant engagement and program effectiveness. This study integrates the DeLone and McLean Information Systems Success Model and the Technology Acceptance Model (TAM) to evaluate system effectiveness and user acceptance. Variables from the DeLone model—System Quality, Information Quality, and Service Quality—were incorporated into TAM constructs, influencing Perceived Ease of Use (PEOU), Perceived Usefulness (PU), and User Satisfaction, which shape Attitude Toward Use (ATU) and Behavioral Intention (BI).
Survey data from 342 DTS administrators were analyzed using Structural Equation Modeling (SEM). Results show PEOU significantly influenced PU (path coefficient = 0.900) and ATU (0.594), while PU and ATU collectively explained 70.3% of BI. High PEOU (0.91) and PU (0.87) scores highlight the importance of system usability and utility. However, low ATU stems from organizational misalignment, as the system’s focus on participant quality contrasts with the DTS priority on participant numbers. Addressing this misalignment, enhancing system features, and improving service quality can boost adoption and foster a digitally skilled workforce aligned with Indonesia’s evolving demands.