Artificial Intelligence as Supplementary Learning Tool in Pharmacology Tutorials: Students’ Perceptions and Knowledge Outcomes 

Technical Summary: Artificial Intelligence as a Supplementary Learning Tool in Pharmacology Tutorials

Problem Statement
While Artificial Intelligence (AI) is increasingly integrated into medical education for personalized learning and simulation, there is a scarcity of discipline-specific evidence regarding its efficacy in pharmacology tutorials. Pharmacology presents unique challenges due to its high volume of factual content, complex mechanisms, and the necessity of integrating pharmacokinetics and pharmacodynamics. Although active learning strategies are recommended, literature has largely focused on problem-based or case-based learning rather than AI-assisted tutorial support. Furthermore, evidence from Southeast Asian undergraduate medical settings is particularly limited. Concerns persist regarding the reliability of AI-generated information, curriculum alignment, and the risk of uncritical reliance reducing independent reasoning. This study addresses the gap by evaluating preclinical medical students' perceptions of AI-assisted learning in respiratory pharmacology and descriptively examining knowledge outcomes.

Methodology
The study employed a quantitative quasi-experimental design conducted at Universiti Kuala Lumpur Royal College of Medicine Perak (UniKL RCMP) during the 2024–2025 academic year.

• Participants: Preclinical MBBS students enrolled in pharmacology sessions. A total of 69 students completed the perception survey (100% completion rate), while 106 and 50 students completed the pre-test and post-test, respectively.
• Intervention: Freely available AI tools, including Chat Generative Pre-trained Transformer (ChatGPT), AI summarizers, quiz makers, flashcard generators, and slide assistants, were introduced as adjuncts to conventional respiratory pharmacology tutorials. These tools were used for concept clarification, revision, self-directed learning, and question generation, not as replacements for formal lectures.
• Data Collection:
  • Knowledge Assessment: Pre-test and post-test multiple-choice questions (MCQs) focused on core respiratory pharmacology concepts were administered via the institutional virtual learning environment.
  • Perception Survey: A validated, structured questionnaire covering five thematic domains (perceived usefulness, engagement/motivation, learning autonomy, usability/confidence, and overall satisfaction) was used. Responses were recorded on a 5-point Likert scale. The instrument demonstrated excellent content validity (I-CVI = 1.00).
• Statistical Analysis: Descriptive statistics (means, standard deviations, frequencies, percentages) were used. Inferential statistical testing was not performed due to unequal pre-test/post-test sample sizes and a "ceiling effect" where baseline knowledge scores were already very high.

Key Results

• Knowledge Outcomes: Baseline knowledge was exceptionally high, with a mean pre-test score of 9.61/10. The post-test mean score was 9.52/10. No meaningful descriptive improvement was observed, attributed to the high baseline performance of the cohort and the resulting ceiling effect.
• Student Perceptions: Overall perceptions of AI-assisted learning were positive (mean score 4.15/5).
  • Highest Rated Benefits: Improved understanding of pharmacology topics (4.43/5), time-saving during self-study (4.36/5), more effective doubt clarification (4.31/5), and willingness to recommend AI for other subjects (4.30/5).
  • Usage Patterns: Students most frequently used AI to explain difficult concepts and summarize lecture content. ChatGPT was identified as the most helpful tool.
  • Concerns: The lowest-rated domain was "Usability," specifically regarding confidence in the relevance and accuracy of AI-generated content (mean 3.77/5). "Incorrect or outdated information" was cited as the most prominent practical challenge.
• Future Intent: 90.47% of respondents indicated they would continue using AI tools in future studies. The primary motivations were time-saving and study efficiency (63.49%), followed by revision and practice question generation (26.98%).

Key Contributions

• Empirical Evidence in Pharmacology: The study provides specific data on how preclinical medical students in a Southeast Asian context perceive and utilize AI tools within a pharmacology curriculum, a domain where such evidence is sparse.
• Validation of Supplementary Model: The findings support a model where AI serves as a supplementary resource rather than a replacement for traditional teaching. Students valued AI for conceptual support and independent study but recognized the necessity of faculty oversight.
• Identification of Barriers: The research highlights that while engagement and efficiency are high, concerns regarding the accuracy and currency of AI-generated content (hallucinations) remain a significant barrier to full trust, particularly in a clinically consequential subject like pharmacology.
• Methodological Insight: The study documents the "ceiling effect" in knowledge assessment when baseline performance is high, suggesting that future studies in similar high-performing cohorts require more discriminative assessments or longitudinal designs to measure cognitive gains.

Significance and Claims
The paper claims that AI tools are well-accepted as supplementary learning resources in pharmacology education, valued specifically for clarifying complex concepts and supporting self-directed learning. However, the authors modestly conclude that measurable knowledge gains could not be demonstrated due to methodological constraints (high baseline scores and unmatched groups).

The significance of the study lies in its recommendation that AI should be integrated cautiously as a complement to conventional teaching, with active faculty oversight to ensure content accuracy and curricular alignment. The authors argue that while students are eager to adopt these tools for efficiency, educators must guide students in the critical appraisal of AI-generated content to prevent superficial learning. The study concludes that future research should employ matched designs, discriminative assessments, and longitudinal follow-up to rigorously evaluate the educational impact of AI-assisted pharmacology learning.