Cambridge Mathematics asks what it means for AI to “do maths”
An education-focused guide to separating correct answers, reasoning, representation, and mathematical understanding.
Visual references
3 images
The classroom question behind the headline
Cambridge Mathematics examines a question that is easy to oversimplify: does an AI system do mathematics, or does it predict plausible mathematical language? The answer depends on the task. A model may solve a familiar exercise, fail on a small variation, or produce a correct result with an explanation that does not justify it.
A better classroom vocabulary
Ask four separate questions: Is the answer correct? Is the method valid? Can the learner explain the representation? Can the result be checked independently? This turns a vague AI debate into a teachable assessment framework.
Practical activity
Give students one problem, one AI solution, and one deliberately altered problem. Ask them to audit the assumptions, test a special case, and improve the explanation. The goal is mathematical agency, not faster copying.