Faculty — The Stony Brook School; integrated Claude for AI argument reasoning and refinement (ongoing as of Aug 2022).
Integrated the Claude (Anthropic) large language model into a 9th grade Humanities course to teach logical reasoning and Socratic argumentation. Students interrogate, rebut, and refine AI-generated arguments, effectively performing iterative review and improvement of model-produced text. The course emphasized evaluating argument quality and encouraging students to revise outputs in response to critiques. • Used Socratic discussion to structure critique of AI-generated arguments. • Required rebuttals and refinement steps to improve reasoning. • Applied evaluation across philosophy, theology, literature, and political thought topics. • Positioned student feedback and revisions as a form of training signal aligned with RLHF-style improvement loops.