AI in the classroom: Research focuses on technology rather than the needs of young people
Peer-Reviewed Publication
Updates every hour. Last Updated: 9-Dec-2025 13:11 ET (9-Dec-2025 18:11 GMT/UTC)
A new study from the University of Würzburg's Chair of Mathematics Education shows that AI research for STEM education focuses too much on technology and neglects the holistic development of students.
A new study reveals how our brains store and change memories. Researchers investigated episodic memory - the kind of memory we use to recall personal experiences like a birthday party or holiday.
They showed that memories aren’t just stored like files in a computer. Instead, they’re made up of different parts. And while some are active and easy to recall, others stay hidden until something triggers them.
Importantly, the review shows that for something to count as a real memory, it must be linked to a real event from the past. But even then, the memory we recall might not be a perfect copy. It can include extra details from our general knowledge, past experiences, or even the situation we’re in when we remember it.
The team say their work has important implications for mental health, education, and legal settings where memory plays a key role.
Mathematical reasoning is a fundamental aspect of intelligence, encompassing a spectrum from basic arithmetic to intricate problem-solving. Recent investigations into the mathematical abilities of large language models (LLMs) have yielded inconsistent and incomplete assessments. In response, we introduce MathEval, a comprehensive benchmark designed to methodically evaluate the mathematical problem-solving proficiency of LLMs in various contexts, adaptation strategies, and evaluation metrics. MathEval consolidates 22 distinct datasets, encompassing a broad spectrum of mathematical disciplines, languages (including English and Chinese), and problem categories (ranging from arithmetic and competitive mathematics to higher mathematics), with varying degrees of difficulty from elementary to advanced. To address the complexity of mathematical reasoning outputs and adapt to diverse models and prompts, we employ GPT-4 as an automated pipeline for answer extraction and comparison. Additionally, we trained a publicly available DeepSeek-LLM-7B-Base model using GPT-4 results, enabling precise answer validation without requiring GPT-4 access. To mitigate potential test data contamination and truly gauge progress, MathEval incorporates an annually refreshed set of problems from the latest Chinese National College Entrance Examination (Gaokao-2023, Gaokao-2024), thereby benchmarking genuine advancements in mathematical problem solving skills.