Why do we need Artificial Intelligence in assessment?

The Artificial Intelligence (AI) education revolution is coming. Or at least, that’s the view of Anthony Seldon in his book that investigates whether AI can be the linchpin that enables a move towards a radical new system of education (Seldon, 2018). Seldon defines five unsolvable systemic issues in our current education system:

1. Failure to defeat entrenched social immobility

2. Inflexible progress through the education system

3. Teaching overwhelmed by administration

4. Large class sizes inhibit personalised and breadth of learning

5. Homogenisation and lack of individuation of personality

These five issues largely convene around the ability to personalise education for each individual student all of the time, and it’s no wonder that this is a problem for teachers. The time and effort it would take to investigate everyone’s learning preferences, career ambitions, current knowledge, skills and behaviours − and by proxy, the gaps in these areas, the reasons for these gaps, and to finally understand the level of support for learning available outside of school - and only then plan a personalised journey through a bespoke curriculum, appears to be an impossible task. Enter assessment powered by AI.

Imagine a school, college, training provider or university where each student was given a personal learning coach to guide them through their education journey. A coach who learns about an individual’s current abilities and their future aspirations. A coach who can deliver small, low stakes, adapted assessment at optimised intervals to celebrate learning and identify gaps and misconceptions in skills, knowledge and behaviour. A coach who could suggest an adapted learner journey through the curriculum for each individual student, using assessment responses to suggest the most optimal sequence through the content for maximum depth of learning and personal development, in the minimum time.

This is the potential for AI in assessment. This powerful digital coach would enable (human) teachers to provide a personalised and adaptive learning experience for every student based on AI-powered assessments and feedback. Teachers would be freed from progress monitoring and instead, get to spend more time ensuring learning is wide and deep. Teachers could take the time needed to build deeper relationships with those currently disadvantaged by the system, enabling more people to use learning and knowledge to access previously unachievable opportunities. Schools, colleges, and universities could become places of enormous human diversity, with students developing in a truly individual and holistic manner through their learning journey.

There is much to do before this is a reality. AI systems need to be trained by expert teachers to replicate excellent pedagogical decisions across the full range of subjects. Ethical considerations on the use of AI need wide discussion resulting in agreed frameworks for use (The Institute for Ethical AI in Education, 2021), processes that protect the student and the quality of their education (Ross, 2021) and avoid the potential for baked-in inequality reported in other industries (Alonso, Kothari and Rehman, 2020). Teachers, students, parents and the wider sector need to change their mindset on what education looks like, starting with the removal of the teacher from the front of a physical class. However, these challenges are not impossible and in fact, there are already many AI-assisted assessment solutions in use, such as adaptive learning systems, AI-assisted marking and feedback, AI-assisted assessment creation and AI plagiarism checks (Jisc, 2021 and Ross, 2021). The AI education revolution has already begun.

A good example is the use of an AI-assisted learning and assessment platform at Basingstoke College of Technology (BCoT) which provides learners with personalised learning content following two diagnostic assessments. The maths team at BCoT are refining their approaches all the time and have had the most success with adult learners, who have benefitted from personalised independent learning and instant feedback on their knowledge checks. Younger learners who need to be stretched and challenged have also benefitted from the approach – however, others have learned better when the teacher intervenes to override the AI decisions and plan a path through the content that is concurrent with the classroom topics. In the US, AltSchools plunged headlong into a fully personalised learning approach with mixed success. Here, the collection of for-profit private schools struggled to fully transform themselves into the tech schools of the future before many closed down (Adams, 2019). There are lessons to be learned from both examples in the deployment of the technology - in particular how radical the changes are to the traditional classroom model - and how to best use human teachers in the AI-assisted model.

At NCFE, our Assessment Innovation Fund is actively looking for ideas to enhance the future of assessment and many of the fund’s applicants are using assessments powered by AI in their work. Examples include the use of AI to mark learner work, to provide instant feedback and next steps advice, and to provide feedback on a learner’s online activity. The common thread running through these applications is how best to deploy AI to support the work of educators to add value to the learning experience.

If these early adopters can iron out the wrinkles in AI-assisted teaching, we can expect education systems to scale up rapidly to a place where AI personalised learning, powered by AI assessment, is the norm for most students - facilitated by great teachers adding the human touch to learning, which looks to be as needed in the future as it has been in the past.

References

Adams, S (2019) Can AltSchool – the edtech startup with $174m from billionaires like Zuckerberg and Thiel – save itself from failure? Available at: https://www.forbes.com/sites/susanadams/2019/01/30/can-altschoolthe-edtech-startup-with-174m-from-billionaires-like-zuckerberg-and-thielsave-itself-from-failure/?sh=6bcd59b91997. Accessed 06/09/21.

Alonso,C. Kothari,S. Rehman,S. (2020) How Artificial Intelligence Could Widen the Gap Between Rich and Poor Nations. Available at: https://blogs.imf.org/2020/12/02/how-artificial-intelligence-could-widen-the-gap-between-rich-and-poor-nations. Accessed 09/12/21.

Jisc (2021) AI in Tertiary Education: A summary of the current state of play. Available from: https://www.jisc.ac.uk/reports/ai-in-tertiary-education. Accessed 01/09/21.

Ross, J (2021) Does the rise of AI spell the end of education? Available at: https://www.timeshighereducation.com/features/does-rise-ai-spell-end-education. Accessed 06/09/21.

Seldon, A (2018) The Fourth Education Revolution: Will artificial intelligence liberate or infantilise humanity? The University of Buckingham Press: UK.

The Institute for Ethical AI in Education (2021) The Ethical Framework for AI in Education. Available at https://www.buckingham.ac.uk/research-the-institute-for-ethical-ai-in-education/. Accessed 01/09/21.