What Can AI Really Do for Learning?
The Stanford Framework in Action
Well, here’s something AI can do. Take the Stanford paper establishing the Human Agency Scale (HAS) and apply it to learning tasks and strategies as described in some of my favorite “how to learn” documents and by learning science in general.
Here's Claude’s analysis based on cognitive science principles:
Student Learning Tasks
H2 (AI needs minimal human input for optimal performance)
Spaced Practice Scheduling: AI excels at optimizing spacing intervals and reminding students when to review material, but humans must do the actual studying.
H3 (AI and human form equal partnership)
Retrieval Practice Tools: AI can generate practice questions and track performance, while humans do the cognitive work of retrieval.
Pomodoro Technique: AI handles timing and reminders; humans do the focused work.
Interleaving Schedules: AI can optimize the mixing of topics; humans do the learning.
H4 (AI requires human input to successfully complete)
Creating Brain-Links Through Practice: AI can provide problems and track progress, but humans must make the cognitive connections.
Concrete Examples: AI can suggest examples, but humans must understand how they relate to abstract concepts.
Dual Coding: AI can help find visuals, but humans must create meaningful word-visual connections.
Testing and Teaching Others: AI can facilitate connections and provide platforms, but humans do the explaining and learning.
Prioritization ("Eat Frogs First"): AI can help identify task difficulty, but humans must make contextual judgments.
H5 (Essential human involvement)
Focused/Diffuse Mode Switching: Requires metacognitive awareness and intentional mental state management.
Elaboration: Demands deep understanding and personal meaning-making to explain concepts in detail.
Exercise: Physical activity that must be performed by humans.
Memory Palaces and Metaphors: Requires personal creativity and meaningful associations.
Active Learning Outside Class: Needs human curiosity, initiative, and self-direction.
Teacher Tasks
H3 (Equal partnership)
Assessment and Progress Tracking: AI can handle data analysis and pattern recognition; teachers interpret and act on insights.
H4 (AI requires human input to successfully complete)
Classroom Organization: AI can suggest organizational tools and systems, but teachers must implement based on student needs and classroom dynamics.
Cognitive Support and Instruction: AI can provide resources and adaptive content, but teachers must understand individual student needs and provide appropriate scaffolding.
Identifying and Removing Distracting Cues: AI could help identify potential distractions, but teachers must make contextual decisions about classroom management.
H5 (Essential human involvement)
Providing Emotional Support and Warmth: Requires human empathy, relationship-building, and emotional attunement.
Teaching Self-Regulation Skills: Needs human modeling, coaching, and responsive support based on individual student emotional states.
Managing Student Negative Emotions: Requires emotional intelligence, cultural sensitivity, and relationship-based interventions.
Creating Psychologically Safe Environment: Demands human understanding of social dynamics and individual student needs.
Meaningful Feedback on Learning: While AI can grade, meaningful feedback that advances learning requires human understanding of student thinking processes.
Key Insights
Why Most Learning Tasks Require High Human Agency (H4-H5):
Metacognitive Nature: Effective learning strategies like elaboration and self-monitoring require human self-awareness and intentional cognitive control.
Meaning-Making: Learning involves creating personal, meaningful connections that require human understanding and creativity.
Social-Emotional Factors: The research shows that emotional support, relationship quality, and psychological safety are crucial for learning—all requiring human emotional intelligence.
Transfer and Generalization: Moving from specific examples to general principles requires human insight and pattern recognition.
Contextual Adaptation: Both learning and teaching require constant adaptation to specific contexts, individual differences, and emerging situations.
Where AI Can Be Most Helpful (H2-H3):
Logistical and administrative tasks
Optimization of spacing and scheduling
Pattern recognition in large datasets
Providing practice opportunities and resources
Tracking progress and identifying trends
This analysis suggests that while AI can provide valuable support for learning and teaching, the core cognitive and emotional work of education requires substantial human agency and cannot be effectively automated.