MIT researcher explains why AI models still make mistakes
On October 18, 2025, Neil Thompson of MIT’s CSAIL lab spoke at the NDTV Summit about the real reasons behind AI model errors and their limitations. He emphasized cleaner data, stricter testing, and growing compute costs as key factors for improvement. Benefit: Gives the public and policymakers a realistic view of AI’s current boundaries. Significance: Encourages smarter investment and regulation focused on transparency and data quality.