21 Ethical Use of AI
With the increasing prevalence and usability of AI, it might be tempting to employ it within the communication of mathematics. However, it is worth noting the several downfalls of AI that is raising concerns about its applicability within the discipline of mathematics. As Professor Ulrike Tillmann, Director of the Isaac Newton Institute for Mathematical Sciences says, “AI is rapidly changing mathematics. From training and teaching to formalisation and publication, AI has the potential to reshape every aspect of our work. As both individuals and institutions, it’s important to think through the consequences of AI adoption and agree as a profession how we want to mitigate the risks posed by AI.” (Isaac Newton Institute for Mathematical Sciences 2026)
As such, it is important to ensure that new technologies support mathematics as a discipline grounded in proof and trust. In considering AI as a tool in education, it is important to consider it with the objective of transforming the understanding of learners by changing the process of learning, rather than transforming the products of education - for instance, the worksheet, or the lesson plan (Singh 2026).
21.1 Pitfalls
Several pitfalls of AI within the communication of mathematics are addressed in the following, with corresponding guidelines and good practices (Almond 2026):
Inaccuracies and Hallucination
The problem: AI chatbots can often confidently get mathematics wrong, despite giving verbose or seemingly well-thought out answers.
Suggested Solution: Do not rely on unchecked output. Use AI to prompt clarification for method explanation or concept clarification to aid mathematical intuition for enhanced communication, instead of just asking for a final answer. Encourage pupils to reflect on their methodology, and encourage questions and discussions.
Over-Reliance
The problem: AI can quickly generate responses but often lacks pedagogical structure, including and addressing misconceptions. Excessive reliance on AI would hence lead to decrease in thinking and problem-solving, hindering the development of mathematical thinking and communication.
Suggested Solution: Use AI for inspiration or ideas, but build core resources tailored to learners. Tailored AI resources should requiring significant editing and structuring on the communicator’s behalf.
Data Privacy Concerns
The problem: Many AI models produce results without citing the human work they build on, or are built on data obtained by violating copyright protections.
Suggested Solution: Disclose tool use, and adhere to principles of copyright and data protection regulation - only upload resources to chatbots if permission has been given.