Technological advancement is a major challenge facing veterinary education. Artificial intelligence (AI) is revolutionising veterinary medicine in imaging, disease diagnosis and telemedicine (Westgate, 2017).
Opportunities for leveraging AI in veterinary education are vast; however, risks exist.
What are the challenges facing veterinary education?
- The cap on UK university fees increased from £1,000 in 2005 to £9,250 in 2017. Rising costs are a growing barrier to veterinary education and contribute to skills shortages.
- Rapid and frequent advancements in veterinary medicine, such as new technologies, treatments and diseases, require the veterinary curriculum to evolve to stay up to date.
- A shortage of qualified veterinary faculty staff to teach veterinary students in some areas.
- The high-stress environment of veterinary medicine can lead to mental health problems, such as depression, anxiety and burnout, among students and faculty staff.
What is AI?
“Artificial intelligence” is a term that was first described in 1955 by John McCarthy, an assistant professor at Dartmouth College. McCarthy defined AI as: “Making a machine behave in ways that would be called intelligent if a human were so behaving” (Cope et al, 2020).
The Oxford English Dictionary defines AI as: “The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages”.
AI is a large field incorporating different technologies and methods, of which machine learning and deep learning are the most common forms. Both use algorithms (problem-solving instructions) and models (representations of systems/processes) to process data and make predictions. For example, a forecasting model may use data about temperature, humidity and air pressure to predict if it will rain tomorrow. The learning aspect of the system could take the forecast, combine it with a learned pattern of activities and send a notification to inform whether to take an umbrella on a walk to work.
The recognition of patterns in data and subsequent prediction is the basis of machine learning. Large amounts of data are fed into an algorithm, which learns from the input and refines the output over time. As an example, an algorithm can be shown a series of images of domestic cats (and told those images are of cats) and then shown images of non-cats (again, informed that the images are not of cats). As the algorithm correctly identifies more cats and is shown more verified images of cats, its ability to positively identify cats improves – hence, the machine can “learn”.
Machine learning is being used in this way within farm animal science to identify livestock systems, such as herd type, with the aim of aiding surveillance for endemic cattle diseases (Brock et al, 2021).
Deep learning is advanced machine learning that uses artificial neural networks that mimic human brain processing – comprising layers of connected nodes known as “neurons”.
Deep learning has the advantage that it can learn from very large datasets to perform complex tasks, such as image and speech recognition, as well as natural language processing. Everyday examples include voice assistance on mobile devices and autonomous vehicle controls.
How is AI being used?
AI features heavily in our day-to-day lives, often without us even knowing it. Simply unlocking your mobile phone with facial recognition is a form of AI.
Businesses are leveraging AI to improve customer service and save money, such as through website virtual assistants (“chatbots”) that resolve customer queries.
Navigation websites and applications use AI to help predict the fastest routes and shortest journey times for their users, based on historical and live traffic and incident data. Rapid advances in AI are resulting in its growing application in medicine. AI has the potential to transform the way health care professionals approach diagnosis and treatment (see Briganti and Le Moine, 2020). For instance, it is helping pathologists and diagnostic imagers to streamline cancer cases (Coccia, 2020).
AI systems can, therefore, assist medical practitioners in reaching a more rapid diagnosis, and by predicting outcomes and best treatment options, which may lead to improved and personalised patient care.
At present, AI applications within the field of veterinary medicine are largely in the research domain, but are expected to be transformative (El Idrissi et al, 2021).
However, AI is beginning to emerge in the clinical setting and within telemedicine. Examples include use within diagnostic imaging, such as the automated analysis of radiographs (Joslyn and Alexander, 2022; Cornell College of Veterinary Medicine, 2023) and the identification and classification of disorders from medical images (Leary and Basran, 2022).
In addition to aiding imaging, AI is assisting the prediction and diagnosis of disease, such as the AI algorithm developed to detect canine hypoadrenocorticism (Reagan et al, 2020).
How is AI being integrated into higher education?
The primary uses of AI in medical education are support to learning, provision of individualised feedback, and aiding the review of curricula.
A review conducted by Chan and Zary (2019) examined the broad uses of AI in medical education, including evaluation of laparoscopic skills, training and evaluation of hand-washing techniques, and analysis and feedback of surgical skills.
Virtual teaching assistants and smart enrolment counsellor chatbots are examples of where AI has benefited higher education. Ashok Goel, a professor at the Georgia Institute of Technology, created a virtual assistant system named Jill Watson. With a 97% success rate, Jill Watson promptly responded to large numbers of student queries and provided the human teaching assistants time to focus on face-to-face learning (Goel and Polepeddi, 2018).
At Georgia State University, staff found that many students enrolling in spring dropped out before course commencement. Pounce, a smart text messaging service (chatbot), was used to answer student queries involving admissions and funding. Dropout rates were reduced by 22% as Pounce assisted students with enrolment tasks for their courses (Page and Gehlbach, 2017). Other examples of AI tools that can be used within higher education include ChatGPT, Explain Like I’m Five, Quizgecko and Trinka. The tools can be used to generate text responses to questions, undertake proofreading and assist with writing, as well as to help teachers quickly generate quizzes, and to help researchers work more efficiently.
Opportunities for AI in veterinary education
Administrative tasks
AI can automate administrative tasks such as grading, generating lecture content and assessments, and communicating with students, as well as identifying and compiling literature.
Using AI to perform administrative tasks allows faculty staff to be more efficient and dedicate more time to high quality in-person teaching, as well as research.
Enhanced learning
AI systems can analyse individual strengths and weaknesses to provide a customised and adaptive learning plan (Kernot, 2016). The learning progress of every student can be tracked, and the content adjusted to provide an adaptive learning experience.
AI chatbots can help students formulate ideas for tasks such as essay writing. Students may benefit from AI-generated learning tools, such as simulators and virtual reality environments (Kernot, 2017), that allow them to practise surgical or client-interaction scenarios in controlled conditions. Such personalised, practical and immersive learning is likely to improve animal welfare.
Finally, teaching AI offers the opportunity to also teach data bias, privacy and security.
Evaluation of teaching, learning, and assessment processes
Feedback from students and other stakeholders can be used to continually improve curricula, while large datasets can be used to identify patterns in student performance.
Increased accessibility and sustainability of teaching materials
Remote learning can be enabled by AI-powered platforms through access to educational material and interactive learning at the click of a button, allowing a cost-effective and flexible approach to continuing professional development.
Avignon et al (2020) suggested that AI can be used where practical training opportunities are limited; for example, due to lack of live animals or carcases during a disease outbreak. A similar approach may help manage learning during a national lockdown.
Further contributing to sustainability and welfare are the ways in which AI can assist in the replacement, refinement and reduction of live animals used for training purposes (Avignon et al, 2020).
Collaboration
AI-powered collaboration tools can allow faculty staff and students to work together more effectively, regardless of their physical location or their discipline.
Innovation
AI provides opportunities for students to develop and test new technologies and tools that enable more rapid veterinary medicine advancement.
AI-powered clinical decision support systems can be used to aid faster and more accurate diagnoses and treatment plans.
Improve veterinary professional numbers
Improved access to educational content, enhanced teaching methods and better collaboration with faculty staff may improve the retention and development of veterinary professionals.
Improved well-being
Efficient teaching and enhanced learning offered by AI may improve the well-being and motivation of students and teachers, by reducing working hours and increasing productivity (Walsh et al, 2019).
Improved affordability
Students’ learning preferences and performance histories can be used to create personalised learning pathways, and help them find the most relevant and effective learning resources for their individual needs and learning styles.
Challenges of AI in veterinary education
Chan and Zary (2019) identified technical difficulties and assessment of the effectiveness of AI systems as the main challenges of AI implementation within medical education.
Other challenges for AI implementation in veterinary education include the following.
Potential for inaccurate output
Some AI systems can be confidently wrong. For example, the generation of false references that look remarkably genuine.
The information recalled by some chatbots is only as reliable as the source.
High costs and accessibility
Training (Grunhut et al, 2022) and implementation costs necessary to adopt AI systems in education may be prohibitive.
A reliable and strong internet connection is required for some AI tools.
Data privacy and security
Medical records and imaging data need to be kept secure.
Information entry into AI systems risks sharing personally identifiable or commercially sensitive information with unknown parties.
Maintenance
AI systems must be developed, maintained and used by skilled staff that may be in short supply.
Limited data availability
The accuracy and effectiveness of AI systems rely on the availability and relevance of high-quality data to test and train the systems.
The inputs can be limited or incomplete in veterinary medicine, and some existing systems only include information up to a certain timeframe.
Reliance on AI systems may result in errors and malpractice – especially if results are biased and/or difficult to interpret.
Quality assurance concerns
Johnson and Labruyère (2023) have shown that a lack of standardised protocols and regulations for testing and quality assurance of AI tools may lead to compromised animal welfare.
Resistance to change
Integration of AI with traditional learning methods will require adjustment time for both students and faculty staff. Lack of awareness, limited resources, and uncertainty surrounding the benefits (and potential risks) of AI may cause hesitancy or barriers to the adoption of AI.
Dishonesty and inhibited learning
The potential use of AI-powered systems by students to generate answers for assessments is dishonest, may be difficult to detect and, where used, may inhibit learning.
Over-reliance on AI
Overuse of AI systems may result in reduced hands-on training, and veterinary professionals who lack both day one competencies and critical thinking skills. Limited exposure to animals and their welfare needs may result in a lack of empathy towards animals (Avignon et al, 2020).
Ethical and legal considerations
The lack of regulation of AI needs to be carefully monitored and considered before its adoption into veterinary education and practice to protect animal welfare, and to comply with laws such as General Data Protection Regulation (Cohen and Gordon, 2022). AI systems must be fully ready and validated before release on to the veterinary market (Webb, 2022).
A basic framework for the implementation of AI in veterinary education
Drawing on the recommendations of the Cornell College of Veterinary Medicine (2023), a framework for implementing AI in veterinary education is shown in Table 1.
| Table 1. Framework for implementing artificial intelligence (AI) in veterinary education | |
|---|---|
| Stage | Consideration |
| 1. Goal setting | Consider areas where AI could enhance veterinary education and choose appropriate AI systems. Goals may include reducing costs, increasing accessibility or raising key performance indicators related to student experience. |
| 2. Model development | Understand and collect the data required to develop models. Establish criteria to assess the performance of AI systems. |
| 3. Training | Adequate training of staff and students to aid understanding of the opportunities and challenges. |
| 4. Continual improvement | Continual adaptation of algorithms and maintenance of data quality to ensure relevance to evolving standards of care and patient needs. |
Could AI make educators obsolete?
A common concern of integrating AI into society is the perception of displacement of human roles. AI is not a panacea and the role of the human remains imperative – not only for monitoring AI systems, but for the provision of human empathy, which is essential in medicine (Montemayor et al, 2022).
AI should not be considered a threat to teachers. Instead, a hybrid model where AI systems and educators work in a symbiotic relationship could be adopted to maximise its benefits (Molenaar, 2022).
AI offers opportunities to transform education, and make it better for students and staff.
Conclusion
Although the veterinary field appears to be in the very early stages of AI adoption, when compared to its medical counterpart, the veterinary students of today will enter a workplace where AI is embedded, whether it be in a clinical, academic or industrial setting.
To succeed, veterinary students must be equipped with the skills to use and critically evaluate AI systems and their outputs. The risks and opportunities of AI should be taught as part of the curriculum.
The integration of AI in veterinary education offers the potential to transform education in many ways. Opportunities exist that could benefit both staff and students.
AI systems can optimise student learning experiences, enhance student outcomes by delivering personal and adaptive learning, and advance the field of veterinary medicine. AI systems can also remove the burden of administrative tasks, and allow staff to focus on more meaningful cognitive interactions and counselling of students.
Use of AI systems in veterinary education has the potential to improve the quality of care and animal welfare. Education and training must be focused on existing veterinarians to ensure they are well-equipped to use AI technologies in practice.
While the opportunities offered by integrating AI into veterinary education are vast, several challenges must be addressed to ensure the responsible, ethical and legal use of these technologies. Challenges such as data security, quality assurance, limited data availability and the potential of over-reliance on these systems need to be considered.
Veterinary students and professionals must be given sufficient hands-on training to ensure competency.
AI is not a panacea; for its benefits to be maximised, we should develop hybrid models where we use AI and humans to their maximum potential, and therefore, “enhance theoretical and practical training, instead of replacing it” (Avignon et al, 2020).
Addressing the challenges will require timely collaboration between industry experts, regulators and educators.
Source : https://www.vettimes.co.uk/article/artificial-intelligence-in-vet-education-risks-and-opportunities/
