Abstract:

This study examines the integration of artificial intelligence (AI) in comprehensive academic tasks in higher education (HE), focusing on an experience conducted in the Financial System course at University of Murcia. The activity analyzed was designed to replicate real-world academic and professional practices by requiring students, organized in groups, to complete a multifaceted project that included information retrieval, content structuring, written report preparation, visual presentation design, and oral delivery. This holistic approach provides an ideal context for assessing how AI supports different stages of academic work and for identifying its benefits, limitations, and pedagogical implications.

Before the public presentation, the groups completed a structured questionnaire requesting information about the use of AI, the tools employed, the purposes of use, the estimated percentage of work completed by AI, and their evaluation of the experience. The questionnaire also explored broader perceptions of AI in HE, including its usefulness, reliability, and desirability of incorporating AI training into degree programs. A total of 54 students participated.

The findings reveal universal adoption of AI for script preparation (100% of respondents) and lower adoption for presentation design (46.3%). On average, students reported that AI contributed 41.5% of the work for the script and 45.6% for the presentation. These figures indicate that AI was widely used as a significant support tool rather than a complete substitute for human effort.

ChatGPT emerged as the dominant tool, used by 98.1% of respondents, primarily for structuring content, retrieving information, and drafting text. Copilots ranked second (42.6%). Canva, used for presentation design, achieved the highest average rating (4.1 out of 5), although only 22.2% of respondents used it. Other tools, such as Deepseek, Gemini, Perplexity, and Dall-e, were used less frequently, with Dall-e receiving the lowest rating (1.6). Overall, students valued AI for its ability to save time, facilitate organization, and enhance presentation quality while recognizing limitations such as superficial content, occasional inaccuracies, and the need for constant verification.

Qualitative feedback reinforced these findings. The students emphasized that AI served as a "starting point", requiring human intervention for refinement and validation. Interestingly, analysis of the performance data revealed that moderate AI use (26–50%) was associated with the highest grades, suggesting that optimal learning outcomes result from strategic integration of AI rather than excessive reliance on it.

Perceptions of AI in HE were generally positive: students rated its usefulness at 4.4 out of 5 and expressed strong intentions to increase future use (mean score 4.0). However, reliability received a lower rating (3.2). Students favored incorporating AI into course activities (mean score of 4.2) rather than creating a dedicated AI subject (3.7).

In summary, this study shows that AI tools play a valuable role in enhancing efficiency, creativity, and engagement in HE when used strategically and critically. The results underscore that how AI is used matters more than how much it is used. Therefore, it is necessary to have pedagogical frameworks in place that ensure that students develop not only the technical competence but also the critical thinking skills necessary to validate and contextualize AI-generated content.

Keywords:

Artificial intelligence, digital tools, higher education, student perceptions, academic workflow.