AI Tutoring Platforms Provided Incorrect Math and Science Answers to Students

In early 2025, multiple AI-powered tutoring platforms faced scrutiny after research documented significant error rates in their responses to student questions in mathematics and science. The platforms affected included Khan Academy's Khanmigo, Chegg's AI tutoring service, and several other ed-tech companies that had rapidly deployed large language model-based tutoring systems to compete in the growing online education market. Educators and researchers began documenting instances where these AI tutors provided fundamentally incorrect answers to problems ranging from basic algebra to advanced calculus and chemistry. The errors were not limited to minor computational mistakes but included conceptual misunderstandings and flawed reasoning processes that could mislead students about fundamental principles. Teachers reported receiving student work that reflected these incorrect methodologies, leading to classroom confusion and the need for remedial instruction. The issue gained prominence when education researchers published studies quantifying error rates across different platforms and subject areas. The research revealed that while AI tutors excelled at providing quick responses and maintaining engaging conversations with students, they frequently failed at the core educational task of providing accurate information. Mathematics problems involving multi-step reasoning and science questions requiring application of principles to novel scenarios showed particularly high error rates. The incident sparked broader debates about accountability in educational technology and the appropriate use of AI in learning environments. Education advocates argued that the rush to deploy AI tutoring systems prioritized engagement and cost reduction over educational accuracy and student outcomes. The platforms involved faced pressure from educators, parents, and school districts to implement stronger quality control measures and provide transparency about their AI systems' limitations and error rates.