New research shows how artificial intelligence (AI) could maybe help address challenges like maths anxiety by using a student’s inputs and identifying signs of anxiety or disengagement during learning.
Published in npj Science of Learning, the Adelaide University study suggests that when AI systems are designed to use the right data and goals, they can adapt their responses to help counteract negative emotional experiences associated with maths, before these feelings escalate.
Lead researcher Dr Florence Gabriel says AI has the potential to transform how maths anxiety is supported, by offering timely, tailored interventions that step through learning and build student wellbeing.
“Maths anxiety is an emotional response characterised by fear, tension, and apprehension when a student is faced with a mathematical problem or test,” Gabriel, a senior research fellow at the university’s Centre for Change and Complexity in Learning, says.
“In some cases, it can be so paralysing that it limits a student’s learning and performance.
“While it’s normal to feel some level of anxiety when encountering challenging subjects, excessive maths anxiety can lead to avoidance, reduced self-confidence and a loss of control – even long-term aversion to mathematical learning.
Gabriel says tailored AI models have the potential to change the way students engage with maths.
“By helping students set realistic, motivating goals aligned with their individual capabilities, and by responding with encouragement when signs of frustration appear, AI can help students feel more competent, motivated and in control of their learning,” she suggests.
The research proposes a new model of mathematics learning where emotional development is treated as central to the design of AI rather than secondary.
Co-researcher Dr John Kennedy says more effort needs to be put into the development and refining of AI models to ensure they are better suited to the realities of education.
Key recommendations suggest AI could support learning in a range of ways.
Learning activities can be tailored by adjusting the difficulty of maths tasks in real time to balance challenge and success, while emotionally intelligent feedback might be provided with the recognition of patterns of frustration or disengagement and responses then shaped in in constructive, personalised ways.
Student autonomy can also be supported by enabling goal-setting and personalised learning pathways that give students greater control, while teachers might benefit from AI offering real-time insights to support more targeted emotional and instructional interventions for students who need it most.
Research suggests that more than a third of adults and children experience maths anxiety.
Indeed, those with the greatest maths anxiety can perform almost four years behind those with the lower levels of maths anxiety.
Co-researcher Dr John Kennedy says more effort needs to be put into the development and refining of AI models to ensure they are better suited to the realities of education.
“Current AI models are trained to provide users with answers they’re happy with, but this can bypass the cognitive processes of learning,” Kennedy, a senior lecturer within the university’s College of Education, Behavioural and Social Sciences, says.
“When students rely on tools that simply generate answers, they only learn how to prompt the system rather than how to think through a problem.
Kennedy says it is imperative that we move beyond this basic use of AI and towards tools designed from the ground up for education – tools that understand local contexts, diverse learning goals and the emotional dimensions of learning.
“This requires a shift in the way researchers work: away from asking what AI can do for educators, and towards asking how educators can shape AI for the benefit of all learners,” he says.
Effective educational AI should not only break problems into simpler steps but also tailor the type of hints it gives and the emotional tone of its responses to support positive attitudes to learning, Kennedy explains.
“That might include recognising delays in responses, deleted text, or patterns of hesitation during problem-solving - but this requires a different approach to training the AI to that commonly used today.
“When AI can adapt to a learner’s emotional state as well as their cognitive needs, it brings us closer to truly supportive and intuitive learning tools.”
The research paper titled ‘Pragmatic AI in education and its role in mathematics learning and teaching’, can be accessed here.
