Teaching in Higher Ed Deep Background: Using AI as a Co-Reasoning Partner with Mike Caulfield
34 snips
Oct 2, 2025 Mike Caulfield, a researcher known for the SIFT verification method, dives into the innovative use of AI as a co-reasoning partner. He shares insights about balancing AI's strengths, like source discovery, with its limitations in tracing provenance. Mike emphasizes teaching students to approach AI interactions iteratively, transforming critical thinking into practical doing. He reflects on the importance of maintaining the journey of discovery and advocates for ethical AI usage in communities. Plus, he introduces his Deep Background tool to enhance information literacy.
AI Snips
Chapters
Books
Transcript
Episode notes
Starlight Opera Memories And Sound Of Music
- Mike recalls watching musicals at San Diego's Starlight Opera as a child and freezing when planes flew over.
- He uses a Sound of Music reference to connect musical memory to AI discussion.
LLMs Simulate Reasoning, Not Reliable Facts
- Early LLMs were impressive but unreliable for verifiable factual queries before search integration.
- Mike Caulfield found they simulated argument structure well, which suggested educational uses.
Search+LLM Unlocks Contextualized Reasoning
- Adding web search to LLMs transformed their usefulness for fact-finding and context.
- Caulfield used this combo to have AIs simulate reasoning about search results effectively.
