According to Andrej Karpathy, as discussed on Twitter, the use of AI in education fundamentally alters assessment practices because AI-generated homework is undetectable by current tools. Karpathy asserts that all AI detectors are ineffective and easily bypassed, requiring schools to assume that any out-of-class work could utilize AI (source: x.com/karpathy/status/1992655330002817095). As a result, he recommends shifting the majority of grading to monitored in-class work, ensuring students are evaluated on their independent skills. Karpathy emphasizes the need for students to become proficient in both leveraging AI tools and verifying their own work without AI assistance, drawing a parallel to calculator adoption in math education. This shift presents significant opportunities for edtech companies to develop in-class assessment tools and AI literacy programs, responding to the evolving needs of schools adapting to AI's integration (source: x.com/karpathy/status/1992655330002817095).