Percepciones del Aprendizaje con ChatGPT en la educación Empresarial Universitaria

Introduction

The integration of artificial intelligence (AI) into higher education has driven a significant transformation in educational processes. Its application enables institutions and teachers to optimize repetitive tasks, personalize instruction, and use resources more efficiently, which consolidates AI as a key tool for educational innovation. This technological growth has fostered a sustained increase in research that examines its impact on professional preparation at the university level. Nevertheless, it has also generated debate about the role of AI in the development of students' cognitive and creative skills, since scholars question whether it promotes innovation or encourages excessive automation of thought (Franco-Lazarte, 2024).

In the current university context, the adoption of ChatGPT and other generative AI tools has raised expectations because of their potential to optimize academic productivity and provide immediate academic support. However, systematic reviews show that, despite the substantial increase in studies, empirical evidence on their pedagogical effect remains heterogeneous and, in many cases, insufficient to validate concrete benefits (Zawacki-Richter et al., 2019; Bond et al., 2023). Recent research also indicates that students value their practical usefulness but express reservations about reliability, ethics, and adaptation to the academic profile (Stöhr et al., 2024). In this scenario, researchers need to understand how students in business programs perceive their academic experience with ChatGPT 3.5 in order to identify opportunities and challenges in the development of business competencies with the support of artificial intelligence.

The integration of Generative Artificial Intelligence (GenAI), such as ChatGPT, into higher education has raised concerns about academic integrity and the reliability of content that these systems produce. Although researchers recognize its potential to improve academic written production and critical thought, ethical dilemmas about its use persist. Among business students, Elbaz et al. (2024) show that personal morality and religious ethics influence the acceptance of ChatGPT and attitudes toward academic dishonesty. Sila et al. (2023) and George-Reyes et al. (2025) likewise report uncertainty about the accuracy of information that these systems produce, which underscores the need to strengthen ethical digital literacy and critical thought. In the business field, GenAI appears as a driver of innovation that requires new competencies in communication and management (Zhang & Zhang, 2025; Al-Okaily et al., 2025). Studies in Latin America, however, remain scarce; therefore, this work seeks to explore the perceptions and ethical factors that influence its use among undergraduate students in Peru (Velasquez-Astuhuaman & Libaque-Saenz, 2024).

This study aims to explore how undergraduate students in a business school at a Peruvian university perceive their academic experience with ChatGPT 3.5 as an academic support tool during 2023. Knowledge of how future professionals interpret and adopt GenAI proves essential to identify new digital and ethical competencies that respond to the demands of the current business environment. Within this framework, the research seeks to answer the following question: What is the perception of the use of ChatGPT 3.5 in courses for professional preparation in business?

Methodology

To respond to the proposed objective and maintain coherence with the institutional research lines related to applied knowledge generation, the researchers developed the study within the pragmatic paradigm. This paradigm allowed them to address real-world problems through the integration of diverse data sources and complementary approaches and prioritized the usefulness and applicability of results in specific contexts (Weaver, 2018). Within this framework, the researchers adopted the analytical-synthetic method and a mixed approach, which integrated quantitative and qualitative procedures in a complementary manner. The study followed a case study design, with convergent triangulation and a cross-sectional scope; this design allowed the simultaneous collection and analysis of both types of data, as well as their subsequent integration, to achieve a more complete and robust interpretation of the phenomenon under study (Creswell & Plano Clark, 2017).

The population included 328 students who belonged to the Business Administration program during 2023 at a private Peruvian university. To determine the sample, the researchers applied the finite population formula and used a 95% confidence level and a 5% margin of error, which allowed them to establish a sample size of 178 students. The participants belonged to the last four years of the program and attended seven academic sections, under the guidance of three faculty researchers, who also served as instrument administrators. The sample included 77 men (43%) and 101 women (57%), all regular students.

The study used homogeneous probabilistic sampling because the units of analysis corresponded to students from the Faculty of Business in the last four years of the Business Administration program, all of legal age and with advanced knowledge of business management. The participants took the courses Business Development I, Design Thinking, Innovation for Business, Entrepreneurial Thinking, and Digital Transformation.

The study conceived the research technique as the set of systematic procedures that allow researchers to collect, analyze, and interpret data on a given phenomenon (Arias & Covinos, 2020). The researchers applied a mixed questionnaire, composed of closed-ended and open-ended questions, as the main instrument of the study. The instrument consisted of 12 specific questions: the first five, closed-ended, sought to identify prior knowledge about artificial intelligence; the next three, also closed-ended, explored students' perceptions about competency acquisition according to Bloom's Taxonomy (Anderson & Krathwohl, 2001); and the final four, open-ended, addressed the perception of the academic experience mediated by ChatGPT 3.5. This methodological design ensured, on the one hand, quantitative validity through the statistically calculated sample size and, on the other hand, qualitative depth through category saturation in the analysis of open-ended responses.

The researchers conducted the study during the 2023 academic year, a period in which the university actively promoted the use of AI tools, such as ChatGPT 3.5, for content development and case analysis. The methodological dynamic included comparative case tables that contrasted the results derived from students' critical analysis with the results that ChatGPT 3.5 generated, in order to examine the consistency, applicability, and added value of AI in practical and reflective academic work.

The study defined statistical analysis as the process that allows researchers to organize, analyze, and interpret data in order to identify patterns, trends, and relationships (Creswell, 2003). The researchers developed this analysis through case resolution in groups of up to five students. The scheduled curriculum included the cases, and students addressed them through two solution approaches: the first, which they prepared on the basis of critical analysis; and the second, which AI generated through various prompts. The researchers considered one academic semester sufficient for the preparation of proposals and the comparative analysis, which favored an active and reflective educational process that William Glasser's educational theory associates with greater participation and knowledge retention (Glasser, 1998). The researchers structured the sequence of activities for the final proposal in four phases:

• Formation of groups and assignment of cases according to curricular development.

• Exploration of the use of ChatGPT 3.5.

• Preparation of a double-proposal solution matrix, one proposed by the students and another that ChatGPT 3.5 generated.

• Preparation of a final report with the comparative analysis between the results that ChatGPT 3.5 projected and the results that the student groups obtained.

Finally, the three faculty members involved in the study, who also authored the research, validated the 12-question instrument, the double-proposal solution matrix, and the structure of the final report to guarantee the coherence and relevance of the procedures used. The researchers also obtained informed consent from the students and clearly informed them about the objectives and procedures of the study. The team ensured confidentiality, anonymity, and information security, and established a commitment to share the results with the participants at the end of the research.

Throughout the research, the team respected the fundamental ethical principles of educational research and guaranteed voluntary participation, informed consent, confidentiality, and anonymity of the participants. The researchers clearly informed the students about the objectives, procedures, and scope of the study, as well as the academic use of the information they collected. The team also ensured responsible and secure data management, avoided any use other than that foreseen, and protected the participants' identities. In view of the dual role of the faculty researchers as authors and instrument administrators, they maintained a reflective and transparent attitude throughout the research process to minimize possible bias and safeguard the academic and ethical integrity of the study.

Results

The results below derive from the analysis of a five-question diagnostic test that the researchers applied to students to explore their prior knowledge about the use of artificial intelligence. The researchers expressed quantitative data through frequencies and percentages with an exclusively descriptive character, without any claim of statistical inference or application of significance tests. In this regard, they used percentages as guiding indicators to describe general trends in participants' responses. Tables 1 to 5 organize these results by question, which allows the identification of the students' initial level of familiarity with AI and establishes a reference point to describe their responses after the incorporation of ChatGPT 3.5 into the development of the courses.

The researchers conducted the qualitative analysis of the open-ended responses through an inductive codification process that included successive reviews of the discourses, identification of units of meaning, and subsequent classification into new categories. In this process, the researchers identified recurrent discursive patterns, especially in the categories related to the perceived usefulness of AI and support for case analysis. Most students expressed assessments concerning the support that AI offered for reasoning and the organization of ideas, while several participants highlighted ChatGPT as a complementary resource to contrast their own proposals. These recurrences allowed the researchers to identify shared trends in student perception, and they illustrated them through representative textual quotes, which they selected for their coherence and recurrence in each analyzed category.

The first question of the instrument sought to identify students' prior level of knowledge about the fundamental characteristics of AI at the start of the academic cycle. This analysis allowed the researchers to establish the initial familiarity base with which students approached the courses and provided a reference context to compare the responses that they recorded in later phases of the study.

The initial results show that most students had an intermediate level of knowledge about artificial intelligence, which suggests a sufficient, although uneven, conceptual base to integrate these tools into their academic experience (see Table 1). This finding indicates that, although the group did not begin from a position of absolute unfamiliarity, the students still needed to strengthen their applied understanding of AI in academic contexts.

In relation to the tools that students knew (Table 2), participants recognized and used ChatGPT more than any other platform, far above options such as Bing or Midjourney. This predominance shows the rapid penetration of ChatGPT in the educational field and its direct association with practical learning, which remains consistent with the objective of the study: to analyze students' perceptions of its use in professional preparation. Overall, these results reflect an academic context in which students knew AI tools but still had incipient experience with them, which makes ChatGPT a key resource to observe how perceptions and digital competencies evolve throughout the course.

Table 3

Previous Use of Artificial Intelligence in Student Activities.

Note. Prepared by the authors based on the results of the survey administered to the students (2023).

The results show that most students had not used AI tools in their academic activities before the beginning of the cycle (see Table 3). This lack of practical experience contrasts with the level of theoretical knowledge that the study had previously identified and confirms that AI use in the educational context remained incipient.

Table 4 presents students’ initial perception of AI’s potential to support research and innovation. Fewer than half of the students recognized its usefulness, whereas a considerable proportion expressed indifference or skepticism. This combination of curiosity and uncertainty indicates an early exploratory stage in AI adoption as an academic tool.

However, by the end of the academic cycle (Table 5), students’ perceptions of artificial intelligence had changed. More than 80% of the participants indicated that AI, particularly ChatGPT, supported activities related to research and the design of value propositions. This variation reflects a greater appreciation of the tool after its incorporation into course development. The responses obtained through the instrument show a shift from the initial perception toward a more favorable assessment of its academic usefulness.

Together, the results in Tables 1 to 5 show a shift in students’ responses from an initial level of basic knowledge to greater application and appreciation of artificial intelligence in business education. The variations in response frequencies and the categories derived from activities with ChatGPT support this interpretation. These results describe changes in the way students perceive the usefulness of AI for the development of digital and innovation competencies in business contexts.

The results of the sixth question (see Table 6) show that students mainly valued AI’s contribution to the analysis of research and innovation processes. They also highlighted its capacity to facilitate comprehension and the acquisition of new knowledge. This finding suggests that students perceived ChatGPT not only as a consultation tool but also as active support for the development of higher-order cognitive skills, in line with Bloom’s taxonomy, especially the analysis and evaluation of information.

Regarding professional preparation, the results of the seventh question (Table 7) show that students consider AI a contributor to the development of key business competencies. They emphasize its usefulness for the proposal of innovative ideas and the analysis of business environments. To a lesser extent, they recognize its contribution to the examination of business models or customer profiles, which indicates a preference for ChatGPT use in the early stages of ideation and validation of value propositions.

Table 8 shows that most students perceive artificial intelligence, specifically ChatGPT in this study, as a tool that mainly facilitates comprehension (21%) and analysis (21%) of the research and innovation process. In contrast, only 5% considered the traditional method, without AI support, more effective. This result suggests a positive assessment of AI as a complementary pedagogical tool and aligns with the objective of the study: to examine students’ perceptions of ChatGPT use in business education.

In the qualitative phase, the open-ended questions deepened this perception. Students reported initial technical difficulties, such as “it was hard for me to formulate the questions to the system properly,” and identified clear advantages for academic development, as reflected in the statement “ChatGPT helped me structure my research work better.” Several students also emphasized that, unlike other courses without AI, “the use of ChatGPT made the process more dynamic and understandable.” Finally, most participants expressed interest in the continued use of AI tools because they enhance autonomy and creativity in research.

These results reinforce the idea that the responsible integration of AI into higher education can foster analytical and research skills, which aligns with current trends in educational innovation (Arias & Covinos, 2020).

Figura 1

Figure 1 presents the organization of the categories and subcategories derived from the qualitative analysis. The structure centers on the main variable identified through induction: the AI-based academic experience, understood through students’ trust in its advantages and in the types of support they perceived.

The first category addresses difficulties during AI use. Most students reported no significant obstacles (55.1%), whereas 44.9% stated that they had faced some initial difficulties, mainly related to the formulation of appropriate questions and the interpretation of the system’s responses. This finding reflects a moderate adaptation curve in the use of tools such as ChatGPT, a common process in early adoption contexts of educational technologies.

The second category focuses on the beneficial aspects of AI integration. Students mainly highlighted rapid access to information (50.6%) and prompt responses (33.1%), along with the possibility of contrastive verification of information (16.3%). These results show that students value the immediacy and informational breadth that ChatGPT offers as support for their academic and research processes.

The third category compared the AI-based experience with courses that did not use this technology. The results show a clear tendency toward a positive perception of AI-supported academic work. A total of 60.7% of students considered that ChatGPT favors deeper academic development, whereas 39.3% stated that it accelerates content comprehension. These results confirm that AI integration can improve both the quality and efficiency of academic work and strengthen the pedagogical purpose of its implementation.

Finally, the fourth category addresses students’ interest and attitude toward AI use. The results reveal broad acceptance and trust (73.0%), compared with a smaller group that expressed caution or reservations (27.0%). These figures suggest that, although concerns about reliability and ethics remain, students show a predominantly positive disposition toward the continued incorporation of AI into professional education.

Overall, the quantitative and qualitative results demonstrate that students recognize ChatGPT as a tool that enhances their autonomy and comprehension in business education. This evidence supports the central objective of the study: to analyze students’ perceptions of the impact of AI on university-level education and research processes.

Discussion

The purpose of this study was to answer the following research question: What is undergraduate students’ perception of the use of ChatGPT 3.5 as an academic support tool in business courses during 2023 at a Peruvian university? The results show that students expressed a predominantly positive perception of AI use in their academic process. In particular, they highlighted ChatGPT’s contribution to the comprehension, analysis, and application of knowledge related to research and innovation, which confirms its potential as a support tool for the development of higher-order cognitive competencies. Overall, these findings confirm that the pedagogical incorporation of generative AI promotes more active, autonomous, and contextualized academic experiences, consistent with contemporary approaches to competency-based education.

These results agree with Selwyn (2020), who argues that the emergence of technologies such as AI transforms traditional instructional dynamics toward more personalized, adaptive, and student-centered models. Similarly, Vera (2023) recognizes that AI has the potential to strengthen educational processes when institutions integrate it under ethical and pedagogical criteria, which promotes a more reflective and responsible relationship between students and technology. In this sense, the positive perceptions among Peruvian students suggest a conscious and critical acceptance of ChatGPT use, which constitutes a favorable basis for its sustained implementation in university environments.

The findings also coincide with Crawford et al. (2023), who highlight that tools such as ChatGPT can stimulate collaboration, critical thought, and creativity by offering academic environments that encourage exploration and the resolution of real problems. These points of agreement reinforce the idea that AI functions not only as a technological assistant but also as a cognitive mediator that broadens opportunities for meaningful academic development.

Likewise, the coherence between these results and research from other fields supports the cross-cutting nature of AI’s educational impact. For example, Quezada et al. (2022) found similar benefits among Law students, where AI contributed to the development of analytical and argumentative thought. Karan (2023) also reported favorable perceptions in dental education, where students recognized improvements in their academic and innovation processes. Together, these convergences show that confidence and positive disposition toward AI constitute decisive factors for its effective integration across different academic contexts, regardless of discipline.

The results also align with the contributions of Holmes et al. (2019) and Mhlanga (2023), who maintain that the rise of generative language models, such as ChatGPT, transforms university education by promoting a more inclusive, agile, and student-centered model. However, Holmes and Porayska-Pomsta (2023) warn that this integration involves ethical and pedagogical challenges related to academic authorship, information reliability, and the formation of critical judgment. In this regard, students’ perceptions in this study reflect both confidence and caution toward AI use, which shows the coexistence of enthusiasm for its benefits and prudence toward its academic and ethical implications.

A complementary aspect emerges when these findings are compared with those of Nazaretsky et al. (2025), who demonstrated that trust in educational technology directly influences perceived usefulness and willingness to use AI-based tools. This interpretive framework helps explain the experience of Peruvian students, who increased their positive assessment of ChatGPT as the academic cycle progressed and consolidated their trust in the tool as a reliable and effective resource that enhances autonomous academic work.

The study presents several limitations. First, the research took place in a single private university, which limits the generalization of the results to other educational contexts in Peru or Latin America. Second, the cross-sectional design limited the analysis of the long-term effects of AI use in professional education. Third, the study relied on student self-perception as the main source of information, which could introduce cognitive biases in the assessment of their own academic experience.

Future research should expand the study to other faculties and universities, both public and private, to examine how institutional and socioeconomic contexts influence AI adoption in higher education. Future studies should also incorporate a longitudinal approach to observe the evolution of ChatGPT use across different semesters and its sustained impact on the development of analytical, creative, and ethical competencies. Finally, researchers should explore interdisciplinary pedagogical models that integrate AI from a perspective of digital ethics and technological literacy, thereby strengthening the comprehensive and critical education of future business professionals.

Conclusion

The relevance of this study lies in the fact that it constitutes one of the first systematic efforts to analyze, in the Peruvian context, undergraduate students’ perceptions of the integration of ChatGPT 3.5 into higher education, specifically within a business school. The results show that students perceive AI as a tool that enhances active academic engagement, improves the comprehension and analysis of information, and stimulates academic innovation. In this sense, the research provides significant empirical evidence for institutional decisions focused on the design of sustainable pedagogical strategies that effectively integrate AI into university education processes.

In the medium and long term, the findings could contribute to the curricular reconfiguration of business programs because they promote the adoption of methodologies based on human-AI collaboration. This approach would foster the development of digital, analytical, and creative competencies in line with the demands of the global labor market. According to the specific objectives of the study, the integration of ChatGPT not only strengthens cognitive development but also promotes students’ self-confidence, autonomy, and ability to solve complex problems.

Future research should deepen the analysis of the ethical, cognitive, and emotional impacts of AI use in higher education, with attention to the diversity of student profiles and academic disciplines. International comparative studies should also identify good practices for the pedagogical implementation of ChatGPT and other generative tools. Finally, universities should design institutional policies on digital literacy and AI ethics, so that future professionals not only master its technical use but also understand its social, ethical, and human implications, which contributes to the development of a more critical, responsible, and innovative education.

Acknowledgments

We express our deepest gratitude to our family members, friends, and loved ones for their unconditional support and constant encouragement throughout this process. We also extend special recognition to the research team, whose academic diversity and shared commitment brought invaluable richness to the development of this work.

Declaration of Conflicts of Interest

The authors declare that there are no conflicts of interest that could affect the execution of this study, nor do they maintain any personal, academic, or professional relationships that could influence or condition the results obtained and their subsequent interpretation.

Declaration of Funding

The authors declare that they have not received funding from any public, private, or commercial institution for this research. The entire work was carried out independently, using the authors' own resources, ensuring impartiality and scientific rigor at every stage of the research process.