What Does It Mean to Be AI-Literate by Arvind Rao

Arvind Rao’s talk, part of the ‘Kaapi with Kuriosity’ series, introduces the concept of AI literacy through the lens of everyday health technologies. He begins with a live demo of a smartphone app that measures heart rate using photoplethysmography (PPG), illustrating the basic input-model-output paradigm. He emphasizes that many so-called AI applications are actually based on classical signal processing and mathematics, and that the term ‘AI’ is often used as a substitute for understanding. He contrasts the low stakes of AI in entertainment (e.g., Netflix recommendations) with the high stakes in healthcare, where errors can have serious consequences. He discusses issues of reliability, reproducibility, and the impact of device variability on measurements. He highlights examples of AI bias, such as Google’s skin cancer detection tool trained primarily on Caucasian skin, and the risks of applying population-specific measurements to other groups. He also touches on data privacy and the hype surrounding AI, noting that AI has passed many exams but that passing an exam does not equate to real-world competence. He argues that AI literacy is essential for becoming thoughtful, informed users who know when to trust, question, or pause.