Modelling the influence of AI dependence to research productivity among STEM undergraduate students: Case of a state university in the Philippines

John Manuel Buniel ¹ Juancho Intano ¹ Odinah Cuartero ¹ Kenny John Grustan ¹ Roey Sumaoy ¹ Noel Jr. Reyes ¹ Jose Jr. Calipayan ¹ Rhodora Arreo ¹ Dione Duero ¹ Ivilyn Rosil ¹ Shypres Agustin ¹ Trisha Jane Diron ¹ Raiya Jocella Pingol ¹ Jean Vanessa Sapuras ¹ Kimberly Miranda ¹ Jesseca Julve ¹ Marinel Josol ¹ Kyla Rita Mercado ¹ Liziel Latoja ¹ Jasmine Cubillan ¹ Pearl Aj Fallado ¹ Elvie Lyka Duran ¹ Francis Isidore Ambray ¹ Myriflor Miranda ¹ Fae Mylene Etchon ¹ Mia Melody Ramoso ¹ Jehu Roeh Rubenial ¹ Franklin Ganancias ¹ Loth Orozco ¹ Joel Gracia ¹ Necie Notado ¹ Geraldine Darao ¹ Sylvester Cortes ^2*

• ¹ North Eastern Mindanao State University, Tandag, Philippines
• ² Cebu Technological University, Cebu, Philippines

STEM fields—Science, Technology, Engineering, and Mathematics—play crucial roles in advancing knowledge, driving innovation, and addressing challenges by means of several mechanisms including research. Consequently, STEM curricula in higher education institutions prepare undergraduate students taking these fields to engage and produce quality research outputs in preparation for their future careers or roles. The advent of several educational resources help these students to perform research-related tasks including artificial intelligence. Although AI use is viewed as inappropriate in doing scholarly works due to concerns about academic integrity and the fear of losing essential cognitive skills, the growing AI dependence among STEM undergraduate students is inevitable. In this regard, the present study seeks to empirically investigate the influence AI dependence toward students’ research productivity, and the mediating roles of research skills, disposition, and self-efficacy. Through literature review, a structural model was proposed and validated. Initially, a research instrument was developed reflective of the constructs present in the structural model where items were also generated using literature review. Eventually, an online survey was conducted and recorded 834 valid responses from STEM undergraduate students. Results revealed that from seven hypotheses proposed in the structural model, six are supported except the causal path between AI dependence and research productivity. The paths between AI dependence to research skills, dispositions, and self-efficacy are supported as well as the paths between these three to research productivity. This indicates the mediation of research skills, dispositions, and self-efficacy between the causal path linking AI dependence to research productivity. The findings of this study imply that strategic integration of AI resources may foster not only skills development but also research motivation and confidence, which together could enhance students' overall research productivity in STEM fields.