Ace
Short answer: AI is changing education quickly: students use it to explain, summarise and quiz, teachers use it to prepare materials, and universities are rewriting their assessment rules. The research so far points one way: AI helps learning when it makes students think, practise and get feedback, and harms it when it simply does the work for them. What hasn't changed is how memory works, so practice and self-testing still matter most.
What is AI changing in education?
- Help on demand. Students can get an explanation, an example or a quiz at any hour, in plain language, on almost any topic.
- Personal tutoring, at scale. One-to-one tutoring has long been known to be powerful: a classic study found tutored students far outperformed students taught in a regular class (Bloom, 1984). AI is the first serious attempt to offer something like a tutor to everyone.
- Assessment. Take-home essays and online quizzes are easier to complete with AI, so many universities are moving towards in-person exams, oral assessments and assignments that show the process, not just the final answer.
- Teaching. Teachers increasingly use AI to draft questions, examples and feedback, which saves preparation time but still needs checking.
What does the evidence say about AI and learning?
Early research is mixed, and the difference comes down to design. In a large study with high-school maths students, unrestricted access to a chatbot improved scores during practice but hurt performance on a later test without it, because students leaned on it for answers; a version built to give hints rather than answers largely avoided the harm (Bastani et al., 2025). In a university physics course, an AI tutor carefully designed around good teaching principles helped students learn more than an in-class active-learning lesson (Kestin et al., 2025).
Policy bodies have reached similar conclusions. UNESCO's guidance calls for a human-centred approach, with age limits, data protection and teacher oversight (UNESCO, 2023). The US Department of Education's report argues that AI should support, not replace, teachers and that humans should stay "in the loop" (U.S. Department of Education, 2023).
What isn't changing?
How human memory works. Learning still comes from effortful thinking: retrieving information, applying it and getting feedback. Testing yourself remains one of the most effective study methods we know of (Roediger & Karpicke, 2006). AI changes how easy it is to get explanations and practice. It doesn't remove the need to do the practice yourself.
What should students and teachers do?
- Students: use AI to test you, not to answer for you. Give it your own course material, ask it to quiz you, and check its answers against your notes. Our guide on how to study with AI has step-by-step prompts.
- Teachers: design tasks where the thinking shows: drafts, oral explanations, in-class problems. Share tools that make students practise rather than ones that write for them.
- Everyone: follow your institution's AI policy, and treat AI output as a draft to check, not a final source.
Ace is built around the "make students think" side of AI: it turns your own lecture slides and notes into practice questions, and every answer shows the line from your material it came from, so it can be checked. See also ChatGPT vs Gemini vs Claude for studying.
Frequently asked questions
Will AI replace teachers?
The evidence and major policy reports point the other way: AI works best when it supports teachers and is designed around good teaching, not when it replaces human judgement and relationships.
Is AI good or bad for learning?
It depends on how it's used. AI that prompts students to think and practise tends to help; AI that hands over finished answers tends to harm learning, even when grades look better in the moment.
Can universities detect AI use?
Detection tools are unreliable, so many universities focus instead on assessment design and clear rules about when AI is allowed. Always check your own course's policy.
Sources
- Bloom, B. S. (1984). The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational Researcher, 13(6), 4–16.
- Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences, 122(26), e2422633122.
- Kestin, G., Miller, K., Klales, A., Milbourne, T., & Ponti, G. (2025). AI tutoring outperforms in-class active learning: An RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports, 15, 17458.
- UNESCO. (2023). Guidance for generative AI in education and research. UNESCO.
- U.S. Department of Education, Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations.
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
