Generative Artificial Intelligence (Gen AI) in dental education: Opportunities, cautions, and recommendations

The Commission on Dental Accreditation (CODA) standard 2-10 states "Graduates must be competent in the use of critical thinking and problem-solving, including their use in the comprehensive care of patients, scientific inquiry and research methodology.1" Critical thinking can be defined as a cognitive process that requires an individual to analyze a situation, gather, and evaluate information, and then use reasoning skills to make informed judgments or decisions.2 It involves the use of various mental faculties, such as comparison, analysis, anticipation, and judgment to arrive at a well-reasoned and justifiable conclusion.2 There are different approaches to assess critical thinking skills of dental students, including problem-based learning, the Objective Structured Clinical Exam (OSCE), standardized patient activities, and reflective writing.2, 3 For example, problem-based learning involves presenting real-life scenarios and having students think critically to devise solutions. The OSCE assesses a student's ability to apply critical thinking skills in a clinical setting, while standardized patient activities require students to interact with simulated patients and make informed decisions based on presented symptoms.2, 3 Reflective writing is another tool that allows students to document their thought processes and analyze specific situations or topics.3 By utilizing these various evaluation methods, educators can gain a comprehensive understanding of their students' critical thinking abilities. Artificial intelligence (AI) has rapidly evolved, driven by powerful computers, telecommunications, and machine learning (ML).4 Natural Language Processing (NLP), a technique employed by large language models (LLMs), enables human-like responses to text-based queries. Generative AI (GenAI), a subset of NLP, creates new information from instructions and input data.4 In Dentistry, GenAI holds potential for tasks such as patient information management, office tasks, imaging support, clinical decision-making, and scientific writing.4 Accessible via computers, tablets, PCs, and mobile phones, GenAI applications are accessed frequently by a variety of users.4 GenAI applications are used to some extent in dental settings without much oversight.5 Furthermore, no established guidelines exist for critically evaluating, verifying, and reporting the information produced by these applications specially for dental students.6 It has been suggested that Generative AI can enhance the development of critical thinking in dental students by providing them with tools and opportunities that extend their learning beyond traditional methods.4-6 Therefore, this perspective article aims to discuss the opportunities and cautions associated with GenAI in dental education, and to identify the aspects of GenAI that support or impair critical thinking. Furthermore, the article proposes institutional policies for the implementation of GenAI and recommends domains for evaluating students' critical thinking when GenAI applications are used. Educational institutions have recognized the need to design curricula emphasizing the responsible use of General Artificial Intelligence (GenAI). This also includes providing essential knowledge to faculty and students about GenAI and its applications. Faculty members are encouraged to develop course materials that acknowledge the presence of GenAI and promote academic integrity, critical reasoning, independent thinking, and individual creativity. This approach will enable students to remain informed and responsible when using GenAI.7 GenAI fosters metacognitive skills through timely feedback and guidance. It encourages students to reflect on their improvement strategies while training them to develop time management skills and methods to enhance their educational processes.8 GenAI provides immediate feedback, allowing students to practice clinical reasoning and decision-making, and can help prepare students for clinical practice through simulated clinical scenarios.8 Examples are provided in Figure 1. However, reliance on AI tools could diminish the development of critical thinking skills, and students may become overly dependent on AI-generated advice, instead of arriving at informed decisions independently.9 Other negative implications include a lack of literacy among students and faculty in GenAI, lack of critical evaluation regarding AI-generated content, lack of knowledge about how results were achieved and sources used, and potential GenAI misinterpretation of source data producing inaccurate answers.9 Although the use of Gen AI in medical and dental applications is expanding, no existing curriculum guides students' learning in this area.9 Furthermore, the lack of educators trained in AI, high costs of GenAI software, absence of regulatory policies, lack of specified data protection, privacy concerns, and inadequate definitions for liability hinder the use of GenAI in clinical settings.9 Other disadvantages of GenAI include limited ability to cite references, lack of contextual understanding, lack of comprehension of medical jargon, lack of reflection, and the information is limited to that which is already known and it cannot create, invent, or make discoveries.9 Although the authors identified positive and negative aspects of Generative AI that could impact the critical thinking of dental learners, no currently available evidence supports either one, and outcomes of GenAI implementation on critical thinking are needed (Figure 1). Institutions, with the support of educators, should develop policies and detailed lists of best practices for using GenAI in student assignments, such as providing guidelines to students on how to disclose/cite AI-generated content when writing essays and research papers. In addition, there is a need to indicate which specific parts of a project are completed with the assistance of GenAI, and what percentage of GenAI was used in the project. Furthermore, institutions should set limitations for GenAI and establish human control to monitor these technologies. Examples of human control to monitor the use of GenAI applications would be direct supervision by faculty, required faculty approvals, electronic health record forms developed specifically for GenAI applications, and committees that establish regulations for their use in the classroom and clinical settings. When implementing GenAI in education, institutions should consider developing policies to minimize privacy violations and establishing guidelines for online professional conduct. Specific instructions for using GenAI in clinical environments are needed to ensure that GenAI is used in conjunction with clinical experience and evidence-based knowledge. The use of GenAI in clinical decision-making requires oversight, and the role of dental practitioners is crucial in preventing dental complications and reviewing GenAI systems. Adequate information on whether the Food and Drug Administration or another regulator reviewed the GenAI must be provided, including how and when GenAI can be used and whether the GenAI recommendation(s) can be followed or overruled if there is disagreement with the recommendation. While preparing the dental curriculum with GenAI content, educators should structure teaching methodologies and student tasks that demand critical thinking and problem-solving above the capabilities of AI and include mechanisms that ensure transparency in their use of AI.10 Timely syllabus updates and clear objectives are needed, emphasizing the relevance of GenAI in clinical practice and its impact on the profession. Tasks must be revisited, and educators should aim to assign activities demanding high-order thinking, creativity, and teamwork to assess their students' analytical and critical reasoning, the soundness and precision of arguments, and their persuasive capabilities. Finally, the curriculum should emphasize the strengths and weaknesses of GenAI in diagnosis and therapy and, with the educator's guidance, promote the ethical and responsible use of GenAI in academic settings to promote students' critical thinking10 (Figure 2). While conducting a literature review on the evaluation of critical thinking among dental students utilizing GenAI applications, the authors found a surprising absence of references on the subject. Consequently, viewing GenAI as both a tool (a device used in everyday tasks) and an instrument (a precise, sophisticated device requiring specialized knowledge), the authors suggest that the assessment of dental students' critical thinking should revolve around their familiarity with the tool, their proficiency in its use, and the quality of the outcomes derived from its usage. Therefore, we propose that critical thinking skills of dental learners that use GenAI applications be evaluated in three aspects: prompt selection, applications of critical thinking when considering and using Gen AI, and output evaluation. Prompts refer to the information, instructions, orders, or queries given to GenAI to perform a specific task. In order to achieve this, learners must acquire the necessary information, choose the appropriate prompts, select the GenAI application, and then input the prompt. This process requires applying critical thinking skills, such as problem analysis, reflective thinking, reasoning, and decision-making, and finally, reflective writing to input adequate prompts. A prompt for a GenAI text generator, such as ChatGPT, should be evaluated as adequate if it possesses the following characteristics: pertinent, relevant, specific, knowledge-based, detailed, and provides clear contextual instructions. Specifically, prompts for an image generator AI application such as implant planning software, should be evaluated as adequate if a prosthetic plan drives them, is based on an adequate Cone Beam Computed Tomography (CBCT) obtained with a radiographic guide, and contains an appropriate surface scan that has no distortions and enough surface area to allow for correct merging and proper design of a surgical guide. The learners can demonstrate reflective thinking, problem-solving, and effective communication by identifying whether the CBCT is adequate, applying anatomical knowledge to the interpretation of the CBCT and implant planning, identifying errors in the resulting scan, determining the need for redoing the scan, and answering questions from their faculty and peers. Conversely, unsatisfactory application of critical thinking will manifest in prompts that are vague, irrelevant, and not grounded in current knowledge, incorrect evaluation of the anatomy, no detection of errors in the scan, and incapability of answering questions from faculty and peers. GenAI applications are capable of generating text, audio, image, video, sound, and code. To effectively use these applications, learners must understand the various types of GenAI applications available, and choose the one that best serves their needs. Additionally, they must be aware of the limitations of the selected GenAI application, and understand its structure and workflows. By evaluating the current evidence for different GenAI applications, learners can demonstrate critical thinking skills. They can also apply decision-making by selecting the appropriate GenAI, demonstrate reflective thinking by comprehending the structure and functioning of the GenAI, and show problem-solving skills by identifying errors and correcting them. Satisfactory outcomes of critical thinking applied to the use of GenAI will encompass the accurate selection and appropriate utilization of the application, along with the detection and resolution of any errors encountered. Conversely, unsatisfactory utilization of GenAI will involve erroneous selection, user-related errors, and an inability to identify and resolve user issues. The GenAI output can be text, image, video, audio, or code. The learner can choose to accept or reject the output. If learners accept the output, they will be prompted to explain their decision. Learners who reject the output will be encouraged to provide reasons for that decision and propose alternative solutions. If learners request additional information, they will be asked to state what specifics are needed or what needs to be clarified. Next, they must be able to re-formulate the research questions if needed, and gather appropriate data, followed by data processing and analysis of the quality of AI-generated information. Furthermore, the learners need to perform adequate hypothesis testing and critically interpret the results. They will use reflective writing to articulate complex concepts and prepare written and graphic data for different audiences (peers, faculty, patients). If the GenAI data is planned to be used as is, learners must include appropriate citations and sources. Some GenAI applications for patient care are designed to assist in the selection of treatment and the evaluation of treatment outcomes. In this regard, students' clinical reasoning skills are required to determine if the proposed treatment is appropriate to the patient's preferences (patient-centered care), if it is based on the best available evidence, if the proposed treatment can be applied to the patient, and if the patient is informed about the treatment options and potential complications. The learner should also understand the benefits and disadvantages of the proposed treatment and identify when collaboration between specialties is required. Therefore, the evaluation of the output will include evaluating decision-making, critically evaluating available evidence, teamwork, communication skills, reflective thinking, self-assessment, and outcomes assessment. Satisfactory critical thinking applied to output evaluation entails providing well-explained decisions, accurate data analysis, appropriate reformulation of research questions, clear reporting in various formats to peers, faculty, and patients, and proper authorship attribution. Conversely, unsatisfactory critical thinking in output evaluation involves failing to provide explanations for decisions, inaccurate data analysis, improper reformulation of research questions, and neglecting to assign authorship. The following table offers an initial flexible framework to evaluate critical thinking when using GenAI and provides guidance on the types of GenAI applications, the necessary prompts, the resulting outputs, recommended student activities and the type of critical thinking that these activities involve, and a scoring system (Table 1) Text Audio Speech Effective communication Reflective thinking Self-assessment The input is correct and the output aligns with the given prompts Students correctly analyze the soundness of the output deciding whether to accept it, input new prompts, or reject it Student provides appropriate justification for the decision Input is not adequate Students accept an incorrect output Student can't justify the decision taken Text Code Application programming interfaces (APIs) New code, software algorithms Large sets of 2D and 3D Images Text Optical scans Radiographic data Implant planning software Aligners software Decision-making Reflective thinking Reflective writing Problem-solving Students validate prompts Students introduce new prompts Students correct errors in the output, or confirm its accuracy. Students articulate the rationale behind accepting or rejecting the output Accuracy of prompt is not verified Student uses erroneous prompts Students incapable of properly operate the GenAI application No rationale is provided by the student Text from documents Voice Internet data Books Denti.AI Voice (Perio charting) Text Odontogram Illustrations Critical evaluation of available evidence Reflective thinking Reflective writing Self-assessment Students examine the output in relation to supporting evidence Student discern the level of evidence Students identify and acknowledge errors, inputs new prompts Student validates or rejects the output and provides justification Student appropriately cites the source Student do not compare the output with the available evidence Student incapable of identifying errors Validation is not completed Justification is not completed Sources are not referenced by the student GenAI is swiftly permeating various facets of dentistry. As educators, it is crucial that awareness is raised about the diverse range of GenAI applications available to learners, assess their impact on both student learning and patient care, enhance faculty literacy, develop clear usage guidelines, and establish specific objectives, expected outcomes, and precise assessments for the effective integration of GenAI into dental education. To ensure the protection of patient data and student privacy when using open-source GenAI software, several critical precautions are necessary. These include anonymizing data to remove personal identifiers, encrypting data for enhanced security, and implementing robust access controls like strong passwords, biometrics, and two-factor authentication. Additionally, it is crucial to train all users—students, faculty, and staff—on HIPAA compliance and privacy policies related to GenAI use, conduct regular security audits, and obtain explicit patient consent for using patient data in GenAI applications. The influence of GenAI on critical thinking skills among dental learners remains uncertain, and comprehensive outcomes of its integration are needed. The suggested guidelines for incorporating GenAI into dental education can serve as a flexible framework adaptable to the unique requirements of each institution. The authors acknowledge Dr. Carol A. Lefebvre DDS, MS, Professor, and Dean Emerita from the Dental College of Georgia at Augusta University for her guidance and mentorship during the ADEA Leadership Institute Fellowship Program. The authors acknowledge ADEA, the ADEA Leadership Institute directives, advisors, and ADEA staff for organizing and supporting this project. This project is part of the 2023–2024 American Dental Education Association Leadership Institute (ADEA LI) Fellowship Program. The authors received full support from the Deans from their institutions to participate in the ADEA LI program. The authors declare no conflict of interest.