Of all types of educational institutions, open universities (OUs) are arguably the biggest beneficiary of technological advancements.” I made this argument in a recent paper entitled “Will Artificial Intelligence Enable Open Universities to Regain their Past Glory in the 21st Century?” (2024). Even at the beginning, OUs distinguished themselves from correspondence education institutions and campus-based universities by pioneering the use of popular technology of the day to break the iron triangle of access, cost, and quality. This is what has made OUs a welcome addition to the global higher education sector. The success story of leveraging technology to deliver high-quality, low-cost education to people who would otherwise have no access to higher education is undoubtedly a legacy of OUs, making them a favourite of governments of almost all political hues and a source of inspiration for campus-based universities.

The success story of leveraging technology to deliver high-quality, low-cost education to people who would otherwise have no access to higher education is undoubtedly a legacy of OUs

Now, as new technologies emerge, particularly artificial intelligence (AI), the promise of enabling OUs to once again break the iron triangle has re-emerged, becoming a common refrain among those advocating for AI in education. According to the New York-based HolonIQ, a global impact intelligence platform for innovative education technology companies, “Education is perhaps one of the most obvious areas for the application of artificial intelligence, with the potential to improve access, dramatically reduce cost and accelerate learning outcomes.” In a symposium paper entitled “Applications of artificial intelligence in an open and distance learning institution”, two colleagues from an Asian OU argue that AI can benefit OUs “in terms of ensuring quality, improving pedagogical methods as well as enhancing the overall teaching and learning experience”. They also suggest that AI could catalyse “a significant and highly intriguing paradigm shift” and shape “the future of all open and distance learners.”

unless AI can give OUs a significant advantage over conventional universities, its deployment and implementation may prove counter-productive and further disadvantage OUs in the ongoing competition.

OUs around the world have faced increasing competition from conventional universities since the late 1990s, with “the first-mover advantages that Open Universities had undoubtedly enjoyed in the first 25 years” being “substantially eroded”, to quote Alan Tait, former Pro-Vice Chancellor of the Open University, United Kingdom. Therefore, unless AI can give OUs a significant advantage over conventional universities, its deployment and implementation may prove counter-productive and further disadvantage OUs in the ongoing competition.

The recent paper of mine mentioned above reviews research literature on the affordances or (potential) uses of AI for open, distance, and digital education (ODDE) and critiques the review findings by examining their implications for OUs through the lens of the iron triangle of access, cost, and quality. Personalisation and automation are the two most frequent themes in the literature of AI-enabled ODDE. Cost-effectiveness is another theme, suggesting that AI can render education more affordable to people from low-income families, thus facilitating universal access to education. Finally, creating virtual learning environments is also mentioned as a possible contribution by AI to ODDE. With each of these themes, I argue that OUs ought to tread cautiously, as the reality is far more complex than the rhetoric suggests.

With respect to AI-enabled personalisation, it may not enhance the quality of OU education as promised. First, the separation of tutors and students in space and/or time in the OU context allows for asynchronicity, which actually enables OU tutors to scale up personalisation more efficiently than their classroom-based counterparts. Additionally, OU students benefit from greater autonomy in selecting study pathways most suitable to their individual needs.

Second, by personalising the learning process, human tutors tend to cater for more than their students’ cognitive needs and may also address their emotional, psychological, and even social needs. Obviously, this is what AI is incapable of but what OU students need most, as they often feel lonely, demotivated, and disengaged due to the inherent nature of ODDE learning.

Third, personalisation is inherently idiosyncratic for human tutors who can notice their students’ unique traits and make the appropriate interventions accordingly. In contrast, AI relies on pattern recognition and correlational analysis, which standardizes patterns without considering cultural or contextual differences. Additionally, human tutors can be proactive, taking the initiative to intervene when they notice something unusual about a student, whereas AI requires activation or prompting to respond.

Finally, OU students can pace their learning suited to their idiosyncrasies, usually within the timeframe of a course or programme while campus-based students may be more restricted in this aspect, not to mention that the pace that AI suggests or prescribes may not be what the students aspire to.

All of this raises the question of whether AI-enabled personalisation is truly what OU students need most. AI-enabled personalisation presents a paradox, as it may “recast and reduce the act of education into an individualized and non-social activity”, to quote Professor Neil Selwyn and colleagues in their paper “Making Sense of the Digital Automation of Education” (2022). This could exacerbate the lack of interpersonal and emotional interaction – already a frequently cited disadvantage of ODDE. Therefore, we should not buy into what AI can do for OUs in terms of personalisation unless we have robust empirical evidence that it can do better than human tutors and is more cost-effective.

Similarly, automation in ODDE “should not be to reduce learning to a set of canned and standardized procedures that reduce the student agency, but rather to enhance human thinking and augment the learning process”, warned Kyoungwon Seo et al. in their paper “The Impact of Artificial Intelligence on Learner–Instructor Interaction in Online Learning” (2021). The automated function of AI should be leveraged with caution, especially when the activities involve complex critical, creative and/or innovative competencies, or when they are tailored to specific courses, programmes, institutions, or contexts. We should not seek standardisation or consistency at the expense of uniqueness, creativeness and/or innovativeness.

On the other hand, AI-enabled automation may be desirable for back-end administrative tasks such as logistics, finances, human resources, staff services, class scheduling, course management, general student inquiry and so on, if it can achieve cost-effectiveness. However, when it comes to direct interactions with students – such as programme or course selection advising, career advising, or work which has a direct impact on students’ future, such as identifying dropouts and at-risk students, recruiting, and e-proctoring – AI should be used with utmost caution, with backup measures in place in case it goes wrong. As a purely human-to-human enterprise, education, emotionally charged, is more an art and craft than a science while AI, inherently rational, is the opposite. Research shows that even the most seemingly ‘objective’ work of administration and the most trivial routine classroom activities such as roll call are social and relational rather than purely procedural in nature. Furthermore, given that OU education is often to blame for a lack of human interaction, a balance needs to be struck between automation and human labour.

As for virtual learning environments, they are definitely a blessing for OU students, enabling OUs to overcome shortcomings in such aspects as engaging in collaborative learning, learning hands-on skills, doing experiments, and the like. However, in view of the high costs they incur, these affordances are more rhetorical than practical for the majority of OUs.

human tutors can be proactive, taking the initiative to intervene when they notice something unusual about a student, whereas AI requires activation or prompting to respond.

As argued in my critique in the previous issue of inspired, there is solid evidence that AI is not cost-effective. The cost issue is even thornier if we contextualize the use of AI in OUs. The more customized, more complex, more intelligent, and more accurate the uses of AI should be, the more expensive it will be. OUs as qualifications providers must prioritise and uphold academic integrity to ensure the credibility of their qualifications and the trust of their students and other stakeholders. This means that the AI tools they deploy must be customised to cater for the specificity and diversity of their courses and programmes and therefore require sophisticated software, high-level intelligence, and high-accuracy algorithm. They will be more expensive than those for generic purposes, if they really work. On the other hand, many OU students are not financially-advantaged and may not be able to afford the use of AI in their study. Without cost-effectiveness, there is no accessibility to speak of; the less affordable AI-enabled OU education is, the less accessible it is to OU students.

To summarise, we have yet to see evidence that AI can help OUs break the iron triangle of access, cost, and quality. On top of this, the affordances of AI for ODDE are hardly distinguishable from those for campus-based higher education. Therefore, suppose these affordances work, they are equally beneficial to both OUs and campus-based universities, hence not putting OUs significantly ahead of their conventional counterparts as other technologies did in the past.

we should not buy into what AI can do for OUs in terms of personalisation unless we have robust empirical evidence that it can do better than human tutors and is more cost-effective.

AI is not a silver bullet for the challenges faced by OUs although it has a role to play in OU education. To what extent and in what ways it can help OUs break the iron triangle depends on how it is creatively and innovatively leveraged to consolidate their strengths and overcome their weaknesses. It is imperative to explore how AI can be used to catalyse innovative quality-assured and cost-effective ways of OU education so that OUs will remain open as to people, places, methods, and ideas in the 21st century. For OUs to maintain competitiveness, any AI-enabled innovation should centre on the ‘four opens’ with the aim of breaking the iron triangle. Additionally, the adoption of any AI-enabled innovation should be informed by robust evidence of effectiveness.

With so many uncertainties existing, the enormous financial implications, and the fact that South Asia alone is home to 385 million poor people (according to the Global Multidimensional Poverty Index 2022 released by the United Nations Development Programme and Oxford Poverty and Human Development Initiative), the risks may far outweigh the benefits if OUs in Asia rush to pursue full-scale implementation of AI in their teaching, learning, and administration.