Human-AI Collaboration: Outsourcing vs Offloading and the Rise of Co-Produced Cognition
Mon Feb 02 2026
Recording from the Deep Freeze: Craig broadcasts from snow-covered north Louisiana (running on generator and Starlink!), where AI helped him MacGyver a propane tank solution involving ratchet straps, a plastic bucket, and a shop light. Welcome to the wild world of practical AI applications.
Featured TopicsOboe.com: The Future of Self-Directed Learning?
Craig and Rob explore Oboe (oboe.com), a free AI-powered platform that creates customized courses on virtually any topic in minutes. Craig demonstrates by building a course on AI agents, and Rob becomes his first student. The hosts discuss:
How the platform auto-generates quizzes with reasonable multiple-choice options and helpful feedbackThe potential to revolutionize textbook accessibility with low-cost or no-cost alternativesUsing Oboe to supplement existing textbooks (like adding blockchain content to their own textbook)The limitations: shallow sourcing and need for instructor vettingCredit to the AI and I podcast from Every.to (makers of Lex.page) for the discovery
Security First: The Moltbot Warning
Not all that glitters is AI gold. Rob raises important concerns about new tools like Moltbot that can automate processes but may introduce security vulnerabilities. Key takeaway: Educators must apply the same critical thinking they expect from students when evaluating new AI tools for classroom use.
Craig's Three-Stage Hierarchy: A Framework for Human-AI Interaction
The centerpiece discussion introduces Craig's developmental model for understanding how we work with AI:
Cognitive Outsourcing - AI does the task for you (the "easy" but often problematic approach)Cognitive Offloading - AI handles specific components while you maintain controlCo-Produced Cognition - True collaborative thinking that produces outcomes neither human nor AI could achieve alone
Craig shares his experience co-writing with Claude, comparing it to the collaborative process of updating their textbook with co-author Franz. The magic: AI enables 24/7 expert-level collaboration that would be impossible with humans alone.
The Big Idea: This hierarchy should guide our teaching. Rather than telling students to "think critically" (a vague catchall), educators should actively move students from outsourcing toward co-produced cognition, where AI's power truly unlocks.
Geeking Out on Affordances
Craig unpacks how AI is fundamentally "a bundle of affordances" - potential uses that only matter when actualized. Using the metaphor of a rock (hammer? erosion control? weapon? stepladder?), he explains:
The same AI tool can be used to cheat on an assignment or to write a meaningless email nobody will readWhat matters isn't just what AI can do, but which affordances we choose to actualizespan class="ql-ui"...
More
Recording from the Deep Freeze: Craig broadcasts from snow-covered north Louisiana (running on generator and Starlink!), where AI helped him MacGyver a propane tank solution involving ratchet straps, a plastic bucket, and a shop light. Welcome to the wild world of practical AI applications. Featured TopicsOboe.com: The Future of Self-Directed Learning? Craig and Rob explore Oboe (oboe.com), a free AI-powered platform that creates customized courses on virtually any topic in minutes. Craig demonstrates by building a course on AI agents, and Rob becomes his first student. The hosts discuss: How the platform auto-generates quizzes with reasonable multiple-choice options and helpful feedbackThe potential to revolutionize textbook accessibility with low-cost or no-cost alternativesUsing Oboe to supplement existing textbooks (like adding blockchain content to their own textbook)The limitations: shallow sourcing and need for instructor vettingCredit to the AI and I podcast from Every.to (makers of Lex.page) for the discovery Security First: The Moltbot Warning Not all that glitters is AI gold. Rob raises important concerns about new tools like Moltbot that can automate processes but may introduce security vulnerabilities. Key takeaway: Educators must apply the same critical thinking they expect from students when evaluating new AI tools for classroom use. Craig's Three-Stage Hierarchy: A Framework for Human-AI Interaction The centerpiece discussion introduces Craig's developmental model for understanding how we work with AI: Cognitive Outsourcing - AI does the task for you (the "easy" but often problematic approach)Cognitive Offloading - AI handles specific components while you maintain controlCo-Produced Cognition - True collaborative thinking that produces outcomes neither human nor AI could achieve alone Craig shares his experience co-writing with Claude, comparing it to the collaborative process of updating their textbook with co-author Franz. The magic: AI enables 24/7 expert-level collaboration that would be impossible with humans alone. The Big Idea: This hierarchy should guide our teaching. Rather than telling students to "think critically" (a vague catchall), educators should actively move students from outsourcing toward co-produced cognition, where AI's power truly unlocks. Geeking Out on Affordances Craig unpacks how AI is fundamentally "a bundle of affordances" - potential uses that only matter when actualized. Using the metaphor of a rock (hammer? erosion control? weapon? stepladder?), he explains: The same AI tool can be used to cheat on an assignment or to write a meaningless email nobody will readWhat matters isn't just what AI can do, but which affordances we choose to actualizespan class="ql-ui"...