Artificial Intelligence (AI) and machine learning offer universities new ways to address perennial problems that confront teachers, researchers, managers and students. However, despite universities having been at the forefront of research and development in computer science from an academic perspective, generally they have been slower to realise the potential of these technologies in other areas of their operations. While there are some great examples of AI being applied in universities – Australia's Deakin University employing IBM Watson to respond to student enquiries; the University of Derby introducing a system that monitors data that predicts when students might drop out to signal timely interventions; Professor Ashok Goel at Georgia Tech building Jill Watson, an AI teaching assistant which students cannot tell apart from her human colleagues – solutions like these have yet to take off at scale. They also represent only a fraction of the benefits that AI could bring to higher education and university research.
Back in the early 2000s I was working in a large faculty at a Group of 8 University in Australia. My job involved identifying and working with students who were experiencing, or likely to experience difficulties with their studies for any number of reasons. The hope was that our interventions would increase retention and student success, particularly for this ‘at risk’ group.
Cue artificial intelligence. Or as it was, cue labour intensive, manual data mining! Back then I spent my days combing student files and issuing surveys to students (on paper!) to find evidence to support the connections that I understood anecdotally – between student circumstances, behaviours and academic achievement. I then designed interventions which might have a positive impact, combed more student files, issued more student surveys – rinse and repeat.
Fast forward to now and in my consulting work I still see many universities using similar, manual methods to identify students likely to experience difficulties with their studies. Admittedly the data collection is easier now – as surveys are online – and there are promising approaches to data storage, breaking down the need for costly integrations linking multiple datasets. But AI, by crawling internal and external databases and systems, ingesting relevant data and using predictive analytics to design and deliver first tier interventions, could supercharge this process. Couple this with the learning capabilities of the machine, enabling much more accurate prediction, and you see the potential for AI to continually improve the university’s capacity to understand and impact student retention and success.
AI uses computational features, such as natural language processing, optical character recognition and machine learning to mimic human intelligence. Like humans these systems learn by doing, receiving and processing feedback, adapting, and then doing again. Because the data and neural networks which make up the AI ‘brain’ are bespoke to each institution, the sooner a university makes a start on building the systems required, the sooner they will reap the benefits of ordering their data and ‘training’ the machine.
Like all innovations, AI attracts its share of sceptics and detractors – some with legitimate questions to be answered, if these solutions are to gain wide acceptance. Starting small and soon will allow time for rigorous assessment of the feasibility of possible applications, as well as sound change management to ensure the technologies are taken up and accepted by users – be they students, researchers, teachers or professional services.
We are only just starting to understand the potential for this technology in universities. In ideation sessions Nous has run with clients in higher education several exciting user stories for AI and machine learning have emerged. For example:
“As a researcher, imagine being able to identify the most promising avenues for my next piece of research, based on everything in my field that has ever been published“.
An interface capable of natural language processing linked to a curated corpus of knowledge containing all current and past publications of chosen journals would provide a solid foundation for this capability.
“As a teacher, imagine being able to easily design and modify personalised learning experiences for each of the students enrolled in my class, based on their current level of knowledge, skill and ability”.
An intelligent assessment program would allow real-time feedback for students to improve their understanding and skills, as well as highlighting areas of weakness for teaching staff and designing appropriate, bespoke interventions for each student.
“As a student recruiter imagine identifying prospective high achieving students in high schools and making them compelling offers, before they have even thought about applying”.
Publically available datasets ingested into an AI engine with the capability to identify complex patterns would help to develop predictive models, enabling universities to spot the characteristics of a successful student long before they complete their high school exams.
“As a student advisor imagine if I could stop students from failing or dropping out – early and easily”.
An AI engine could be trained to identify students ‘at risk’ or not sufficiently challenged, and intervene before they fail a subject or drop out. Universities already have access to the data required to achieve this.
The solutions described here are not imaginary – with AI and machine learning technology they are possible now.
Get in touch to talk with us about how your university could make the next great leap forward.