Subbarao discusses the importance of systematic evaluation when assessing AI capabilities, comparing it to guiding a child to find the correct answer. He emphasizes the role of human prompting in enhancing AI reasoning capabilities and the need for careful evaluation techniques in studying artificial organisms like LLMs.
