Fuel, Friction, and the Human Element
Making AI Work in Education (without losing what only humans can do)
Our Journey Today
01
Real World
Some quick statements about AI
02
Frameworks First
Build shared mental models: Fuel/Friction, Cognitive Load Theory, Polarity Management, Process vs Product
03
Real Applications
See frameworks in action across special education, data analysis, multilingual support, and admin tasks
04
Integrity by Design
Move beyond detection to clear norms and assessment redesign
05
Implementation
Take action with concrete tools, activities, and a plan
We'll anchor AI in proven frameworks so we can amplify what matters and remove what wastes time. For both teaching and administration.
A Real-World Example
"I just asked it how to recover a document accidentally deleted in MS Word 365, and this is what it said: Press Win + R, type %AppData%\Microsoft\Word and press Enter. And I was able to retrieve my document. Had no idea that I could get it back. Thought you would like to know that I am using it."
Love dad
This demonstrates AI's power to provide instant, practical solutions, transforming everyday frustrations into moments of discovery and productivity.
Common AI Anxieties
The integration of AI into education is met with valid concerns, ranging from existential dread to practical challenges in the classroom:
“AI will be worse than phones and likely the end of education and humanity”
“Morally it is destroying the earth through its energy use”
“My students use AI to get what I want them to do done quickly without learning, and then they do what they want to do”
“I feel like I’m being gaslit; I fully reject the use of AI. We don’t have to use it, we can turn it off”
Framework 1: Understanding Fuel vs Friction
Fuel (propel new ideas)
  • Benefits and value propositions
  • Resources and support
  • Incentives and motivation
  • Demonstrating Proof
Friction (resist new ideas)
  • Inertia: Existing habits and routines
  • Effort: Time and energy required
  • Emotion: Anxiety and uncertainty
  • Reactance: Resistance to change

Most AI adoption fails from unmanaged friction, not lack of fuel. We're great at adding new tools but forget to remove the barriers that prevent use.
Reduce Friction First
Create Freedom to Explore
Foster a culture of exploration by protecting dedicated time for discovery. Replace some meetings with 'tinker time' where staff can experiment freely without judgment or pressure to produce immediate results.
Publish Clear Norms
Lower emotional friction with transparent guidelines. Staff and students need to know exactly what's expected and allowed.
Fit Existing Workflows
Reduce inertia by integrating AI into current processes rather than creating entirely new systems to learn. Brisk , Google Classroom
Ship Tiny Trials
Build confidence through small wins. Success creates new fuel that motivates further adoption and experimentation.
Framework 2: Polarity Management
Breathing serves as a foundational example of polarity management. Like inhaling and exhaling, some essential functions require both opposites working together in dynamic tension.
Inhale
Taking in resources, new ideas, growth
Exhale
Releasing old habits, consolidating learning, reflection
The key insight is simple yet profound: You can't just inhale OR just exhale – you need both in dynamic tension to sustain life. This foundational example demonstrates how some things require both opposites working together.
This same principle applies to managing the inherent tensions in education and AI adoption. This breathing metaphor helps us understand how to manage AI/Human polarities, ensuring we balance innovation with human-centered learning.
Framework 2: AI/Human Polarities
Just like breathing requires both inhaling and exhaling, effective AI integration requires managing the tension between AI capabilities and human strengths.
AI Upsides
  • Efficiency and speed
  • Personalization at scale
  • 24/7 availability
  • Consistent quality
Human Upsides
  • Judgment and wisdom
  • Encouragement and motivation
  • Belonging and connection
  • Ethical reasoning

Warning Signs
Over-AI: Shallow work, reduced critical thinking
Under-AI: Teacher burnout, inefficient processes
The goal isn't to choose AI or humans—it's to manage the tension and get the benefits of both.
Framework 3: Cognitive Load Theory
Sensory Information
Raw input from our environment that must be processed
Working Memory
Limited capacity processing center where active thinking happens
Long-term Memory
Permanent storage including semantic (facts/concepts) and episodic (experiences/events) memory.
All three components interact and affect each other. AI can both help and hinder cognitive load - it can reduce unnecessary mental strain when used well, but creates shortcuts that weaken deep learning if overused.
Design goal: Balance AI support with educational theory - use AI to support learning principles rather than replace them. Just in Case vs Just in Time Learning
CLT in Action: Avoiding Over-Reliance
Encourage Active Engagement
Use reflection, showing work, and peer teaching to keep students actively engaged rather than passively relying on AI.
Design Productive Struggle
Create assignments that encourage students to struggle productively, using AI as a hint tool rather than solution provider
AI as Partner, Not Crutch
Frame AI as supportive tool that helps students along the way rather than doing the thinking for them
Work with human nature's tendency toward shortcuts - design learning that guides students toward meaningful effort and deeper encoding into long-term memory.
Framework 4: The Process vs Product Rule
1
Process (Minimal AI)
  • Brainstorming to think
  • Outlining and planning
  • Rough drafts
  • Problem-solving steps
  • Retrieval practice
2
Product (AI Welcome)
  • Polishing and editing
  • Layout and formatting
  • Translation services
  • Slide templates
  • Executive summaries
Generating vs Revising
The Classroom Rule: Use AI when the product is the goal; minimize AI where the process is the learning.
Learning
Questioning
Thinking
Communicating
Belonging
These are the North Stars for human work in education. AI can buy us time, but it cannot create belonging—that remains our irreplaceable value.

For too long education has been about moving toward productivity, what if it moved toward these North Stars? Just in case vs. just in time (tension).
Application: Special Education
AI - Reducd Friction
  • Draft goals from assessment and IEP data
  • Level reading materials to appropriate complexity
  • Translate parent letters into home languages
  • Generate visual schedules and social stories
Human Protection
Teachers maintain decision-making authority, provide encouragement, and ensure all practice remains authentic and relationship-centered.

CLT in Action:
AI reduces sensory overload (formatting), supports working memory (scaffolds), while teachers ensure deep encoding into long-term semantic and episodic memory through meaningful practice. Universal Design for Learning (UDL)
Application: Data to Dialogue
01
AI Processing
Generate summaries of complex datasets, identify three potential hypotheses, and suggest discussion questions
02
Human Diagnosis
Analyze AI insights, apply contextual knowledge, and form evidence-based conclusions
03
Collaborative Action
Develop If/Then next steps, assign accountability measures, and monitor progress through human judgment
AI reframes data into dialogue, but teams own the decisions and maintain accountability for student outcomes.
This approach balances AI efficiency with deep learning - AI handles information processing while human teams engage in the meaningful cognitive work that builds long-term understanding
Application: Multilingual Classroom
Two-Column Handouts
Create parallel content in home language and English to support comprehension and family engagement
Real-Time Translation
Provide captions and instant translation for verbal instructions and classroom discussions
Vocabulary Glossaries
Generate academic term definitions in multiple languages with visual supports and context examples

Design with CLT in mind: reduce sensory overload for families while ensuring students still engage in the productive struggle needed for deep semantic and episodic memory formation.
Lower extraneous cognitive load for families and English learners so attention can focus on content mastery and community building.
Application: Administrative Productivity
Meeting Summaries
Auto-generate action items, decisions, and follow-up tasks from recorded sessions to maintain momentum
Grant Scaffolds
Create application outlines, compliance checklists, and budget templates to streamline funding processes
Photo-to-Count
Use image recognition for inventory management, attendance tracking, and resource allocation
Reality Check: Detection Limitations

OpenAI retired its AI text classifier in July 2023 due to low accuracy rates. Current detection tools should be treated as signals, not verdicts.
Why Detection Fails
  • High false positive rates
  • Easily circumvented by paraphrasing
  • Biased against non-native speakers
  • Cannot detect human-AI collaboration
Better Approaches
  • Build process evidence over time
  • Focus on learning conversations
  • Require disclosure of AI use
  • Design assignments that show thinking
Don't outsource professional judgment to algorithmic meters. Trust your expertise and engage students in honest dialogue.
Rebalancing the Conversation
From Policing...
"Is this AI cheating?"
"How do we catch them?"
"Ban all AI tools?"
...To Designing
"What's AI's role here?" (Firebase Studio)
"How do we teach responsible use?"
"Where does AI add value?"
Use our Process vs Product framework combined with Cognitive Load Theory to establish clear, learning-centered norms. Require disclosure of AI use rather than playing detection games.
We get farther, faster by clarifying where AI belongs than by playing endless games of whack-a-mole.
Assessment Redesign: Fast Wins
Show Your Work
Require submission of drafts, notes, and revision history alongside final products
Oral Defense
Brief conversations where students explain their thinking and decision-making process
Localized Prompts
Use school-specific scenarios and local context that AI cannot easily reference
Iterative Checkpoints
Multiple submission points that track thinking development over time
AI Use Disclosure
Simple checkbox requiring students to paste prompts, settings, and describe their process
Peer Teaching & Reflection
Students explain concepts to others or reflect on their learning process to deepen encoding into long-term memory
Effective Prompting Framework
01
Subject
Define the AI's role: "You are a speech-language pathologist..." or "You are helping a 5th grade teacher..."
02
Context
Provide essential background: student needs, grade level, learning objectives, available time
03
Intent
State your specific goal: "Create 3 SMART goals..." or "Write a parent-friendly summary..."
04
Constraints
Set boundaries: reading level, format requirements, length limits, accessibility needs
"You are an SLP working with a 7-year-old with articulation delays. Create 3 SMART goals for /r/ sound production. Write a parent-friendly summary at a 5th grade reading level. Output as a table."
Early Warnings & Course Corrections
Polarity mindset: watch both gauges and adjust quickly. Small corrections prevent big problems.
Essential Guardrails
Verify Facts & Sources
AI can hallucinate information. Always double-check claims, statistics, and citations before sharing with students or families
Protect Sensitive Data
Never input student names, IEP details, family information, or confidential records into open AI systems
Accessibility Pass
Review all AI outputs for readability, cultural sensitivity, and compliance with accessibility standards
Teacher Authority
You set the Process/Product line for your classroom. Make it clear, teach it explicitly, and model responsible use
Default stance: Trust, then verify. Model responsible use and professional judgment.
Spotlight: AI Tools for Educators
Explore a selection of AI tools that can enhance teaching, streamline workflows, and foster student engagement.
AI-Powered Assistants
General-purpose AI for research, drafting, and summarizing complex information:
  • NotebookLM: Research assistant for your personal documents.
  • Gemini: Google's versatile AI assistant.
  • ChatGPT: Leading conversational AI for creative and analytical tasks.
Education-Specific Applications
Tools built for classroom integration, often with prebuilt content and workflows:
  • Flint: AI specifically designed for learning environments.
  • Curipod: Interactive presentations with AI-generated content.
  • Brisk Teaching: AI tools integrated directly into your teaching workflow.
Key Insights & Opinions
Process vs Product Rule
This is the cleanest integrity guideline that aligns with Cognitive Load Theory: protect productive struggle, offload extraneous work. It's both research-based and practically actionable.
Design Over Deterrence
Assignment redesign plus disclosure norms consistently outperforms detection arms races. Focus energy on learning design, not policing technology.
Culture Over Tools
When we frame AI as a visible, teachable partner, students learn responsible use. When we frame it as cheating, usage goes underground and learning suffers.
We can leverage AI to automate the mundane, giving humans the time and space to foster community, deepen our thinking, and cultivate meaningful relationships.
AI will show us what we're uniquely good at by revealing what it cannot do. Let's strengthen the human element.
AI Training Resources for Educators
These platforms offer structured learning paths, best practices, and hands-on training specifically designed for educators looking to integrate AI tools effectively in their classrooms.
Questions & Next Steps
What's your biggest takeaway from today?
Which framework will you implement first?
What friction point will you tackle this week?
Thank you for your commitment to thoughtful AI integration that amplifies human potential in education.
Research Sources
Frameworks & Theory
Evidence Base
All frameworks and recommendations are grounded in peer-reviewed research and field-tested implementation strategies.
Activity 1: Fuel & Friction Mapping
Identify Your Fuel
Name one specific task you could amplify with AI this week. What would the benefit be? How would it save time or improve quality?
Remove One Friction Point
What's the biggest barrier preventing you or your team from trying AI? Time? Training? Tech access? Policy confusion?
Design Your Tiny Trial
Plan a small, low-risk experiment: Who will be involved? When will you try it? What will success look like?
Take 3 minutes to jot down your thoughts. Share with a partner if possible.
Activity 2: The Process-Product Spectrum
1
Process-Protected (No/Low AI)
Mark one task where students need to do the thinking work themselves—where the struggle builds the schema
2
Product-Accelerated (AI Welcome)
Mark one task where AI can speed up production without compromising learning outcomes
This exercise helps establish your classroom or team norms. Be specific about what goes where.
Take 3 minutes to place two actual assignments on this spectrum.
Try It Now: Demo Prompts

Optional live demonstration—feel free to follow along or just observe
1
IEP Goals & Parent Summary
"You are a special education teacher. Create 3 SMART goals for reading comprehension for a 4th grader reading at 2nd grade level. Include a parent-friendly summary at 5th grade reading level. Output as a table."
2
Bilingual Family Letter
"Create a two-column parent letter about our upcoming science fair. Left column in English, right column in Spanish. Keep to 6th grade reading level. Include participation guidelines and volunteer opportunities."
Remember: Generate → edit and humanize → add your professional judgment → note disclosure of AI use
Action Plan
01
Pick Your Pilot
Choose 1 product task to offload to AI and 1 process task to protect from AI interference
02
Publish Your Norms
Share your Process vs Product guidelines with students, colleagues, or families
03
Set Verification Habits
Establish routine checks for accuracy, bias, and alignment with learning goals
04
Schedule Share-Back
Plan a 15-minute reflection session next week to discuss what worked and what needs adjustment
Remember: Small wins, shared often, beat big initiatives every time.
Link to Presentation - Thank You