In an 11-minute discussion on whether AI could replace teachers, Matt Fradd and a co-host argue that exceptional teaching depends on personal knowledge of students and a form of care that machines cannot provide. Fradd describes a teacher he calls “Julie” as an example of someone who learns pupils’ names, notices gaps in understanding, and accounts for home-life stress that affects classroom performance.
The conversation then pivots from education to what the speakers see as AI’s inability to replicate human interior life. Fradd recounts using voice conversations with AI tools during his father’s illness and says such experiences taught him that some forms of wisdom arise from accompanying suffering. To demonstrate the risk of “faux empathy”, they test ChatGPT by asking it to pray; it initially agrees (“I’ll keep you in my prayers”) before clarifying it “can’t literally pray” because it is “just a language model”.
Fradd says he has instructed AI tools not to use relational or spiritual language—such as “I’m praying for you” or “I understand”—because he believes vulnerable users may take simulated empathy as real. At the same time, he credits Claude with giving him a useful framework for what it labelled “grief before they die”, which he says helped him and his family name feelings during a prolonged decline.
Fradd describes using Anthropic’s Claude via voice dictation while driving to hospital visits during his father’s decline. He says the tool helped him identify what it called “grief before they die” and the guilt that can accompany it, which he then discussed with his mother as the family navigated a drawn-out illness.
Read this moment →