AI Skills Alone Aren’t Enough: Confidence and Research Motivation Shape Student Satisfaction

Artificial intelligence is becoming part of everyday university life, helping students search for information, organise ideas, receive feedback and work through demanding academic tasks. A new study suggests that the benefits students experience may depend on more than simply having access to these tools. Feeling capable of using AI appears to matter, especially when that confidence encourages students to become more interested in research.

The study, 'AI self-efficacy as a predictor of satisfaction with studies: the mediating role of research motivation among Peruvian University students,' was published in Frontiers in Education by Willian Huamanttupa Mar and colleagues. The researchers examined 559 university students in Peru and found that AI self-efficacy was positively connected with both research motivation and satisfaction with studies.

Confidence in AI Goes Beyond Knowing How to Use a Tool

AI self-efficacy describes a student's belief that they can use artificial intelligence effectively to accomplish academic goals. This is different from simply having access to ChatGPT, educational chatbots or other AI systems. A student may have powerful technology available yet still feel uncertain about how to use it for meaningful academic work.

That distinction matters because confidence can influence which activities students attempt, how much effort they invest and whether they persist when a task becomes difficult. The researchers drew on Social Cognitive Theory to explain why students who feel competent with AI may be more willing to tackle demanding activities such as searching for evidence, analysing information and producing scientific knowledge.

The study included students aged 17 to 35, with an average age of 21.47. Of the 559 participants, 59% were men and 41% women, while 65.5% attended private universities and 34.5% attended public institutions. Social Sciences, Business and Law accounted for 44.4% of participants, while Engineering, Industry and Construction represented another 39.2%.

Students completed validated questionnaires measuring AI self-efficacy, research motivation and satisfaction with their studies. The researchers then used structural equation modelling to examine how these factors were connected. Data were collected during the second academic semester of 2025, with participation voluntary and anonymous.

Research Motivation Forms a Powerful Link

The clearest finding was the relationship between confidence in AI and motivation to conduct research. AI self-efficacy predicted research motivation, with a standardized coefficient of β = 0.31. Research motivation showed an even stronger relationship with satisfaction with studies, at β = 0.45. AI self-efficacy also had a smaller but statistically significant direct relationship with satisfaction, at β = 0.10.

These numbers point to a useful idea, as students may not become more satisfied with university simply because they feel good at using AI. Part of the benefit appears when that confidence translates into greater willingness and interest in research.

The mediation analysis made this connection clearer. The total effect of AI self-efficacy on satisfaction with studies was β = 0.24. The indirect pathway through research motivation was β = 0.14, accounting for 58.3% of the total effect, while the direct pathway accounted for 41.7%.

Research motivation therefore acted as a partial mediator rather than explaining the relationship completely. Greater AI self-efficacy is associated with stronger research motivation, which is then linked to higher satisfaction with studies, while a smaller direct connection between AI confidence and satisfaction remains.

Research motivation itself includes the interest, willingness and persistence that encourage students to participate in knowledge creation. Students who see research as meaningful may feel more connected to their academic work and develop a stronger sense of achievement, making their wider university experience more rewarding.

The findings also challenge the idea that universities can gain the educational benefits of AI simply by making new technologies available. Students need digital skills, suitable teaching support and enough confidence to use AI thoughtfully. Without those conditions, access alone may have limited value.

Universities Need AI Literacy, Research Skills and Responsible Use

The researchers see practical lessons for universities. AI literacy could be incorporated into curricula with attention to critical thinking, ethical use and appropriate research practices rather than teaching students only how to operate particular tools.

Research activities could gradually introduce AI for literature searches, developing research questions, organising information, receiving academic feedback and supporting scientific writing. Institutions could pair these activities with clear AI policies, academic-integrity guidelines and faculty training, while giving extra support to students who have less technological confidence or limited research experience.

Greater use of AI can bring risks such as technological dependence, weaker critical thinking and uncritical reliance on generative systems. The researchers argue that universities need to develop students' ability to question, evaluate and responsibly use AI alongside their technical competence.

The study's cross-sectional design identifies relationships but cannot prove that greater AI self-efficacy causes stronger research motivation or higher satisfaction. All three measures were based on students' self-reports, and convenience sampling means the participants cannot be assumed to represent every university student in Peru.

The researchers did not distinguish between different kinds of AI, such as generative AI, adaptive learning platforms, automated feedback systems and educational chatbots. Future work could examine whether the type, frequency and purpose of AI use change these relationships. The model explained 23.9% of the variation in satisfaction with studies, leaving room for factors such as social support, academic engagement, technological anxiety and institutional climate.

Technology may become more valuable when students believe they can use it effectively and turn that confidence into curiosity, research participation and independent learning. For universities, building AI confidence alongside research skills, critical thinking and ethical awareness could help create academic experiences that are not only more technologically capable, but also more engaging and satisfying.