Looking into the Future of AI in Higher Education
“It is not that it has been tried and been found wanting; it has been found difficult and untried.” — G K Chesterton
AI is not a new development. It is already widely used in various private and public organizations, especially financial services, manufacturing, retail and healthcare.
By comparison, the deployment of AI in higher education is slow, with many institutions reluctant to embrace its full potential. Yet the impact on teaching, learning, assessment and student support is already significant.
Some developments are causing concern — a rise in academic misconduct cases, algorithmic bias in the use of analytics for admission and student support, and inappropriate grading in automated grading systems — but others see the deployment of AI in a positive light.
Let’s explore what’s next for AI in higher education and the issues these developments give rise to.
Seven major developments between now and 2030
Here’s what we can expect between now and 2030:
- More intelligent content creation engines such as ChatGPT, with better access to research materials and personal data used to generate content
- More widespread deployment of intelligent chatbots and digital assistants supporting students from application to graduation
- More widespread deployment of real-time translation engines, enabling classes to be taught in one language but heard in many
- More use of analytics to shape interventions aimed at student retention, completion and improved satisfaction with their experience of college or university
- More deployment of adaptive and personalized learning, using formative assessment tools to ensure every student masters the required knowledge, skills and capabilities in ways that work for them
- More use of automated assessment tools for the design and development of assessments, as well as automated grading
- More use of AI for matching knowledge, skills and capabilities with available employment opportunities
Examples can be found of each of these developments in use at colleges and universities throughout the world. The challenge is that their use is not widespread or at a scale that shows significant impact and encourages widespread adoption.
Seven major challenges
There are real challenges in the deployment and use of AI in higher education. Here are the significant issues:
- Costs – To deploy AI at scale across an institution requires investments not just in technology but in training and people. Although some services appear free to use, they do require significant cloud storage, dedicated staff and support for those who want to use AI for teaching, learning and assessment.
- Technology infrastructure – Many colleges and universities strengthened their IT infrastructure during the pandemic to facilitate the expansion of online and hybrid learning. Some could quickly deploy AI capabilities at scale, but others would struggle to do so.
- Bias – The algorithms used for analytics do not consider students’ nuanced experiences, which can trap low-income and minority students in low-achievement tracks. Bias is also “baked in” to some of the analytic systems deployed.
- Cybersecurity – To be effective, AI systems need access to large datasets of personal information, especially for adaptive and personalized learning and for helpful interventions. The concern is such access via cloud-based systems could increase the risk of security breaches, already a major problem for education institutions.
- Innovation mindset – The key barrier to more widespread use and adoption of AI-enabled supports is the mindset of administrators and instructors. There can be reluctance to change what has worked in the past, and if there are few incentives, supports or role models for an individual or team to support innovation, it’s easier to simply do business as usual.
- Change management – At the institutional level, although there may be an appetite for risk and change, it can be challenging to lead significant change in colleges and universities. The cognitive and behavioural changes needed across the institution for humans to recognize their new AI collaborator are significant. This may explain why deployment and evidence of efficacy in higher education is slow in coming, especially when compared to other sectors.
- Relationships – The key to effective, engaged learning is the relationship between instructor and student and between students in the same course. There is a fear that because AI is so focused on content, assessment and success, the quality of the relationships that make learning work may suffer. AI might get in the way.
Some colleges and universities have responded to these challenges with pilot funding for small-scale experiments using AI for teaching, learning, assessment and student support. Others see partnerships and alliances, some facilitated by Educause (US) or JISC (UK), as a way to make collective progress using some of the AI tools available.
The key to more widespread deployment of AI is to think of it not just as an automation tool or assistant but as a collaborator — as a resource to support teaching, learning, assessment, research, writing and thinking.
And instead of seeing AI as a potential replacement, it should be embraced as a way to improve the work of teachers, administrators and researchers. After all, that was what Alan Turing intended when he first wrote about AI in the 1950s.
