7. Beyond ChatGPT: Critical thinking in the age of AI — School’s In
Best for: Families concerned that a polished project may conceal limited learner decision-making.
Who: Stanford’s Hari Subramonyam with hosts Denise Pope and Dan Schwartz. Published: October 16, 2025. Evidence: Publisher summary and transcript.
Listen and read the transcript
Key takeaways: Subramonyam distinguishes tools that produce finished outputs from tools that support the learner’s thinking. He discusses Script & Shift, designed to help users form and structure ideas, and encourages adults to ask children about their choices and how they used AI. The important issue is participation in the process—not merely whether AI was present. (ed.stanford.edu)
Ask the mentor: “Show me a point where the learner must choose between alternatives before the tool generates the next result.”
Limit: Much of the discussion concerns writing. Applying its principles to coding mentorship is this desk’s interpretation, not evidence that a particular coding assessment has been validated.
8. How Questions Can Transform Student-Centered Learning — Harvard EdCast
Best for: Distinguishing a capable teacher from a technically impressive demonstrator.
Who: Karen Brennan, interviewed by Jill Anderson. Published: March 5, 2026. Evidence: Publisher summary and transcript.
Listen and read the transcript
Key takeaways: Brennan describes project learning that begins with students’ interests and questions while retaining expert guidance and structure. Her account of a classroom as a design studio makes room for experimentation, difficulty and revision. She also discusses AI as a way to expand what learners can imagine and build, rather than simply offload thinking. (gse.harvard.edu)
Ask the mentor: “When my child gets stuck, how will you decide between asking a question, giving a hint, explaining a concept or demonstrating a step?”
Suggested application: During a supervised trial, observe the response to a genuine misunderstanding—not only the planned demonstration.
Limit: This is a pedagogical discussion, not a private-tutor selection study. It does not establish that an entirely unstructured project is appropriate for every learner.
9. The Impact of AI on Children’s Development — Harvard EdCast
Best for: Parents considering younger learners, heavy AI assistance or claims that a chatbot can replace a human mentor.
Who: Ying Xu, interviewed by Jill Anderson. Published: October 2, 2024. Evidence: Publisher summary and transcript.
Listen and read the transcript
Key takeaways: Xu describes learning benefits from deliberately designed AI interactions while explaining why those findings should not be generalized to all chatbot use. She discusses productive struggle, the difference between assisted performance and retained learning, and aspects of human conversation that AI does not fully reproduce. (gse.harvard.edu)
Ask the mentor: “What will you deliberately leave for the learner to work through, and what can they demonstrate without AI afterward?”
Limit: This older episode remains useful for framing the learning question. It should not be used as evidence about the safety or effectiveness of a product released later, or as proof that all children respond similarly.
10. AI and the future of human learning — School’s In
Best for: Technical parents and teenagers deciding whether learning to code still belongs in an AI project.
Who: Christopher Piech and Judith Ellen Fan with Denise Pope and Dan Schwartz. Published: December 3, 2025. Evidence: Publisher summary and live-discussion transcript.
Listen and read the transcript
Key takeaways: Piech describes learners using AI to pursue more ambitious coding work. Fan discusses limitations in AI’s handling of visual representations and multimodal reasoning. The conversation also examines motivation and the continuing value of human interaction in education. (ed.stanford.edu)
Ask the mentor: “Which programming and reasoning skills remain the learner’s responsibility even when AI generates some code?”
Suggested application: Separate the project brief into learning objectives and delivery objectives. A working application may be one delivery objective; explaining its behavior should be a separate learning objective.
Limit: A live university discussion is not a controlled comparison of private mentorship, camps and self-directed study. Do not infer a guaranteed benefit from a particular format.
11. The Science of Raising Humans in the Age of AI — Screen Deep, episode 034
Best for: Parents considering younger-child provision or extensive automation of adult support.
Who: Dana Suskind with host Kris Perry, published by Children and Screens. Date: July 2026; the reviewed page does not specify a day. Evidence: Episode notes and transcript.
Listen and read the transcript
Key takeaways: Suskind asks whether AI creates more space for human connection or replaces it. The discussion emphasizes reciprocal human interaction and offers questions for evaluating technology’s role in family life. Its relevance to mentorship is the distinction between supporting an adult’s work and substituting for the relationship itself. (childrenandscreens.org)
Ask the provider: “Which parts of the service are delivered by the named human mentor, and which are automated?”
Limit: The episode places substantial emphasis on early development and parenting. Its application to an older teenager’s technical project is indirect; it is not an age-by-age AI curriculum.
12. AI in the classroom: Lessons from the field — School’s In
Best for: A busy parent wanting a broad research orientation before choosing deeper episodes.
Who: Hosts Denise Pope and Dan Schwartz, drawing on conversations with Stanford researchers including Dora Demszky, Victor Lee, Hari Subramonyam, Karin Forssell, Judith Ellen Fan, Christopher Piech and Sam Wineburg. Published: September 3, 2026. Evidence: Publisher summary and transcript.
Listen and read the transcript
Key takeaways: This compilation spans feedback for teachers, educational creativity, information literacy and concerns about critical thinking and bias. It presents AI as a set of different educational uses rather than one intervention with a single outcome. (ed.stanford.edu)
Ask the mentor: “Which problem is each AI tool solving in your teaching plan?”
Limit: This is a compilation of earlier conversations, not a new September 2026 study. It overlaps with other selections here; use it as an orientation option, not mandatory additional listening.
13. Decoding Podcast: Inside the Research on AI in STEM Education
Best for: Evidence-oriented parents designing progress reports or reviewing renewal.
Who: Host Marielle Zhang interviews Karen D. Wang. Publisher: Stanford Teaching Commons. Publisher listing: March 19, 2026; this is not a verified recording date. Runtime: 43:13. Evidence: Official episode summary and linked research; the linked transcript was not accessible during this review.
Open the episode and research notes
Key takeaways: Wang discusses AI’s potential to clarify STEM concepts and students’ difficulty using it without bypassing problem-solving. The publisher’s notes recommend explicit learning goals, modeled use and process-based rather than solely outcome-based assessment. The page links the underlying studies, allowing parents to move beyond the interview’s framing. Date listing. (teachingcommons.stanford.edu)
Ask the mentor: “What evidence of reasoning will your progress report include besides screenshots and finished code?”
Limit: The cited work includes college students. Applying its findings to children’s private tuition requires caution; it does not establish outcomes for executive households.
Sources cited on this page
- Beyond ChatGPT: Critical thinking in the age of AI | Stanford Graduate School of Education ↗ed.stanford.edu
- How Questions Can Transform Student-Centered Learning | Harvard Graduate School of Education ↗gse.harvard.edu
- The Impact of AI on Children's Development | Harvard Graduate School of Education ↗gse.harvard.edu
- AI and the future of human learning | Stanford Graduate School of Education ↗ed.stanford.edu
- The Science of Raising Humans in the Age of AI | Dana Suskind, MD - Children and Screens ↗childrenandscreens.org
- AI in the classroom: Lessons from the field | Stanford Graduate School of Education ↗ed.stanford.edu
- Open the episode and research notes ↗teachingcommons.stanford.edu
- Date listing ↗teachingcommons.stanford.edu