Artificial intelligence pioneer Andrew Ng issued a shocking warning at the NDTV World Summit on October 22, 2026, revealing that AI chatbots are causing a 17 percent drop in closed-book exam scores. While the technology successfully boosts immediate homework grades, it severely damages true student learning and long-term knowledge retention.
Citing research from the University of Pennsylvania and Wharton, Ng exposed the hidden danger of relying on generative AI to complete academic assignments. Students who use chatbots to write essays, solve mathematical equations, and generate code might secure higher marks on take-home tasks, but they completely fail to understand the underlying mechanics.
AI chatbots are terrible for learning. When students use AI, their homework scores go up. No surprise, AI does the work for you, but the final closed-book exam scores go down.
Ng warned that compounding this 17 percent chatbot learning loss over years of schooling puts society on a dangerous trajectory to graduate students who know drastically less. However, he emphasized that the fault lies with the current generation of chatbots rather than artificial intelligence itself, urging developers to build systems that actively support learning instead of replacing human effort.
Beyond the classroom, Ng slammed universities globally for teaching outdated curricula that fail to reflect modern workplace demands. He pointed out the alarming reality that students can still graduate with a computer science bachelor’s degree without ever using an AI coding assistant or making an API call to a frontier model.
Highlighting the brutal reality of the modern tech sector, Ng delivered a blunt ultimatum regarding hiring practices. He stated explicitly that he refuses to employ software engineers who lack practical experience working alongside artificial intelligence agents, underscoring an urgent mandate for global educational systems to reform.
TL;DR
- AI pioneer Andrew Ng revealed that students using AI chatbots score 17 percent lower on closed-book assessments.
- Homework scores rise because AI does the work, but true knowledge retention plummets.
- Ng warned this learning loss compounds over years, graduating students who know significantly less.
- Universities face intense pressure to update computer science curricula that still ignore AI coding assistants.
- Ng stated he will not hire software engineers who do not know how to use AI tools.
