Designing Needs- and Attention-Aware AI Learning Tools for Engineering Education: Insights from Psychological Outcomes

This study investigates how AI chatbots influence the psychological needs of undergraduate engineering students, specifically through the lens of Self-Determination Theory (SDT). While much research focuses on cognitive outcomes like grades or task efficiency, this paper examines affective outcomes: autonomy (the sense of agency), competence (the sense of mastery), and relatedness (the sense of connection). The researchers surveyed 206 engineering students to understand how baseline motivational states, self-efficacy, and attention challenges shape the perceived benefits of using AI tools in everyday academic contexts.

The researchers found that AI chatbots are not perceived as a uniform support tool. Instead, students reported that the most significant benefit of AI was the relief of competence frustration—essentially helping them overcome roadblocks that would otherwise stall their progress. Benefits regarding autonomy were moderate, and benefits regarding relatedness were the weakest, suggesting that chatbots currently struggle to foster a sense of social connection or belonging for these students.

Crucially, the study reveals that a student's internal state—specifically their baseline competence frustration, personal agency, and inattention—matters far more than demographic factors like age or gender. The data also indicates that inattention acts as a moderator; students with higher levels of inattention experience the relationship between their baseline needs and the perceived benefits of AI differently than their peers, highlighting the need for more personalized, neurodiversity-affirming design.

As AI becomes ubiquitous in engineering education, these findings suggest that developers should move beyond "one-size-fits-all" tools. By understanding that students primarily use AI to manage frustration and maintain momentum, designers can create tools that better target these specific psychological needs. The results emphasize that for AI to be truly effective, it must be "needs-aware" and "attention-aware," adapting its scaffolding to support students who struggle with self-regulation and focus, rather than simply providing answers.