🎓 UALS Comprehensive Workshop: 3-Hour Training

Workshop Title: "Universal Adaptive Learning System (UALS): A Hands-On Introduction to 100-Agent Intelligent Tutoring"

Duration: 3 hours (with breaks)

Target Audience: Educators, instructional designers, educational technologists, researchers

Format: Comprehensive hands-on training with demos, activities, and discussions

Learning Objectives

By the end of this workshop, participants will be able to:

  1. Understand the three core learning modes (KE, SPL, SBCAT) and when to use each
  2. Experience personalized adaptive learning through hands-on activities
  3. Explain how the 100-agent ITS architecture personalizes learning
  4. Analyze AI recommendations using the workflow visualization tool
  5. Apply UALS to their own educational contexts (curriculum or competency-based)
  6. Evaluate learner progress using xAPI analytics and learner models

Workshop Schedule (3 hours)

Part 1: Introduction & System Overview (30 minutes)

9:00 - 9:30 AM

1.1 Welcome & Icebreaker (5 minutes)

  • Participant introductions
  • Quick poll: "What's your biggest challenge in personalizing learning?"

1.2 The Adaptive Learning Problem (10 minutes)

Three Critical Challenges:

  1. Philosophy rigidity: Traditional systems designed for single pedagogy
  2. AI opacity: Black-box decisions without transparency
  3. ITS fragmentation: Narrow focus (cognitive OR assessment, not both)

Real-World Impact:

  • Students transfer institutions → learning history lost
  • Teachers can't validate AI recommendations
  • Systems can't adapt to diverse learner needs

1.3 UALS Solution Overview (15 minutes)

Four Key Innovations:

  1. Multi-Philosophy Framework:
    • Curriculum-based: Domain → Subdomain → Concept
    • Competency-based: Category → Competency → Proficiency
    • Same UI, different content generation
  2. Three Learning Modes:
    • Knowledge Explorer (exploration)
    • Socratic Playground (inquiry)
    • Scenario-Based CAT (assessment)
  3. 100-Agent ITS:
    • 30 production-ready, 70 extended
    • 17 categories (learner modeling, pedagogy, ethics, neuroscience)
    • 10 coordination levels (4-100 agents, 1-40s)
  4. Explainable AI:
    • "Show AI Thinking" workflow visualization
    • Complete transparency for trust-building
View Dashboard

Part 2: Knowledge Explorer (KE) - Constructivist Exploration (30 minutes)

9:30 - 10:00 AM

2.1 KE Theory & Design (10 minutes)

Pedagogical Foundation:

  • Constructivist learning theory (Piaget, Bruner)
  • Cognitive load theory (Sweller)
  • Interactive concept mapping with progressive disclosure
  • Schema construction through exploration

Key Pedagogical Principles:

  • Learner-driven exploration (not forced sequence)
  • Schema building through visual knowledge structures
  • Just-in-time prerequisite access
  • Curiosity-driven discovery

Agent Coordination (Level 2, 11 agents):

  1. Learner Model → Retrieves knowledge state
  2. Mastery Estimation → Identifies gaps/strengths
  3. Pedagogical Model → Selects exploration depth
  4. Adaptive Sequencing → Recommends next concept
  5. Affective State Monitoring → Assesses confidence
  6. Knowledge Tracing → Predicts trajectory
  7. Intervention Timing → Decides scaffolding needs
  8. Domain Model → Provides concept relationships
  9. Misconception Detection → Identifies errors
  10. Explanation Generation → Adapts complexity
  11. Feedback Generation → Creates encouragement

2.2 Hands-On Activity: Explore Photosynthesis in KE (15 minutes)

Instructions:

  1. Navigate to UALS dashboard
  2. Select "Curriculum-Based Learning"
  3. Choose "Biology → Cellular Processes → Photosynthesis"
  4. Launch Knowledge Explorer
  5. Explore concepts, complete practice questions
  6. Unlock advanced topics by mastering prerequisites

Observation Prompts:

  • How does the system adapt content to your responses?
  • What visual representations help understanding?
  • How does progressive disclosure guide exploration?
Try Knowledge Explorer

2.3 Group Discussion: KE Applications (5 minutes)

  • When would you use KE with your learners?
  • How could KE support different learning styles?
  • What subjects/topics are ideal for KE?

Part 3: Socratic Playground (SPL) - Guided Inquiry (40 minutes)

10:00 - 10:40 AM

3.1 SPL Theory & Design (10 minutes)

Pedagogical Foundation:

  • Socratic method and guided inquiry (Collins & Stevens, 1982)
  • EMT dialogue framework (Expectation-Misconception-Tailored)
  • Productive failure (Kapur, 2008)
  • Zone of Proximal Development (Vygotsky, 1978)

Five Hint Agent Types:

  1. Conceptual: "What principle applies?"
  2. Procedural: "What's the next step?"
  3. Reflection: "Why didn't that work?"
  4. Worked Example: Complete solution
  5. Error Analysis: Systematic mistake identification

Four Interaction Modes:

  • Socratic: Questioning over telling
  • Demonstrative: Worked examples
  • Practice-focused: Repeated application
  • Study Mate: Peer-like conversation

Agent Coordination (Level 3, 20 agents):

All 11 from KE plus Problem Generation, Worked Example Generation, Metacognitive Support, Curriculum Alignment, and Performance Analytics.

3.2 Hands-On Activity: Solve Problems in SPL (20 minutes)

Instructions:

  1. From photosynthesis KE, click "Practice in SPL"
  2. Attempt first problem (plant health scenario):
    • Try solving without hints first
    • Request conceptual hint if stuck
    • Request procedural hint if still stuck
    • View worked example as last resort
  3. Use "Study Mate" mode for conversational support
  4. Complete 2-3 problems with varying difficulty

Observation Prompts:

  • How do hints guide without giving away answers?
  • What happens when you make a mistake?
  • How does the EMT framework address misconceptions?
  • How does difficulty adapt to your performance?
Try Socratic Playground

3.3 Demo: Five Hint Agent Types (5 minutes)

Instructor Demonstration:

  • Show same problem with different hint agents
  • Explain when each agent type is most helpful
  • Demonstrate "Worked Example Agent" for complex problems

3.4 Group Discussion: SPL Applications (5 minutes)

  • How is SPL different from traditional problem sets?
  • When would scaffolding (demonstrative mode) be needed vs. Socratic mode?
  • How could SPL support struggling vs. advanced learners?
☕ BREAK (10 minutes) - 10:40 to 10:50 AM

Part 4: Scenario-Based CAT (SBCAT) - Intelligent Assessment (30 minutes)

10:50 - 11:20 AM

4.1 SBCAT Theory & Design (10 minutes)

Pedagogical Foundation:

  • Item Response Theory (IRT) and adaptive testing
  • ITS-inspired assessment (Hu et al., 2023): Learning during testing
  • 2PL model: Difficulty (b) and discrimination (a) parameters
  • Adaptive algorithm: θ estimation with maximum information

Key Assessment Principles:

  • Efficiency: 40-60% fewer items than fixed-length tests
  • Precision: Standard error < 0.3 in 8-12 items
  • Formative: Feedback and hints available during test
  • Diagnostic: Identifies specific strengths and gaps

IRT 2PL Model:

P(correct) = 1 / (1 + e^(-a(θ - b)))
Where θ = learner ability, b = item difficulty, a = item discrimination

4.2 Hands-On Activity: Take SBCAT Assessment (15 minutes)

Instructions:

  1. From SPL, click "Assess Mastery with SBCAT"
  2. Complete adaptive assessment (8-12 items)
  3. Observe difficulty adaptation after each item
  4. Request hints if stuck (note: penalty to ability estimate)
  5. Review final diagnostic report:
    • Final ability estimate (θ with confidence interval)
    • Mastery level (novice/intermediate/advanced)
    • Strengths identified
    • Gaps for remediation
    • Recommended next steps

Observation Prompts:

  • How does question difficulty change based on your responses?
  • What feedback do you receive after each item?
  • How does the final report identify strengths and gaps?
  • What recommendations are provided for further learning?
Try Adaptive Assessment

4.3 Group Discussion: SBCAT Applications (5 minutes)

  • How is SBCAT different from traditional quizzes/exams?
  • When would you use SBCAT: pre-assessment, formative, summative?
  • How could diagnostic reports inform instruction?
  • What are the ethical considerations of adaptive assessment?

Part 5: The 100-Agent ITS Architecture (40 minutes)

11:20 AM - 12:00 PM

5.1 Agent Architecture Overview (10 minutes)

100 Agents Across 17 Categories:

  • Core Learner Modeling (4)
  • Tutoring Interaction (11)
  • Content Generation (17)
  • Pedagogical Strategies (15)
  • Assessment (11)
  • Cognitive Support (10)
  • Social-Emotional Learning (9)
  • Social & Collaborative (6)
  • Domain-Specific (6)
  • Accessibility & Inclusion (3)
  • Teacher Support (4)
  • Analytics & Research (7)
  • AI Ethics & Fairness (4)
  • Longitudinal Learning (4)
  • Real-World Learning (4)
  • Parent & Community (3)
  • System Intelligence (2)

10 Coordination Levels:

Level Name Agents Time Use Case
1Lightning41-2sQuick checks
2Quick112-4sRecommended
3Standard204-8sComprehensive tutoring
5Professional5012-18sAdvanced analysis
10Complete10025-40sFull ecosystem

5.2 Demo: "Show AI Thinking" Workflow Visualization (15 minutes)

Instructor Demonstration:

Step 1: Launch Learner Onboarding

  1. Click "Get AI Recommendation" from dashboard
  2. Enter learning goal: "Understand photosynthesis"
  3. Select Analysis Level 2 (Quick, 11 agents, 2-4s)
  4. Submit

Step 2: View Recommendation Results

  • System recommends: "Socratic Playground (SPL)"
  • Rationale: "You have foundational knowledge but need application practice"
  • Click "🧠 Show AI Thinking" button

Step 3: Explore Workflow Timeline

  • Visual timeline shows 11 agent invocations
  • Color-coded events (agent invoked, workflow started, completed)
  • Playback controls (play/pause, speed: 0.5x, 1x, 2x, 4x)
  • Click individual agents to see their reasoning

Key Teaching Points:

  • Transparency: See exactly which agents were consulted
  • Reasoning: Understand why SPL was recommended
  • Trust: Educators can validate AI decisions
  • Research: Complete audit trail for analysis
Try AI Recommendation

5.3 Hands-On Activity: Try Different Analysis Levels (10 minutes)

Instructions:

  1. Return to Learner Onboarding
  2. Try Analysis Level 1 (Lightning, 4 agents) - Notice speed
  3. Try Analysis Level 3 (Standard, 20 agents) - Notice depth
  4. Try Analysis Level 5 (Professional, 50 agents) - Notice comprehensiveness
  5. Compare workflow visualizations across levels

Observation Prompts:

  • How does recommendation quality change with analysis level?
  • What's the tradeoff between speed and depth?
  • Which level for rapid in-session recommendations vs. high-stakes decisions?

5.4 Group Discussion: Agent Coordination (5 minutes)

  • How does explainability build trust in AI recommendations?
  • What insights can educators gain from workflow visualizations?
  • How could 10-level coordination support different use cases?
🍽️ LUNCH BREAK (30 minutes) - 12:00 to 12:30 PM

Part 6: Multi-Philosophy Framework (30 minutes)

12:30 - 1:00 PM

6.1 Philosophy-Agnostic Architecture (10 minutes)

Two Educational Philosophies:

  • Curriculum-Based Learning: Domain → Subdomain → Concept
  • Competency-Based Learning: Category → Domain-Specific Competency → Proficiency Level
Universal Template Principle: "ALL philosophies MUST use identical templates, UI, and functionality. Only content differs."

Backend DRY Principle: Single implementation for all philosophies

6.2 Hands-On Activity: Compare Philosophies (15 minutes)

Part A: Curriculum Mode

  1. Select "Curriculum-Based Learning"
  2. Navigate: Science → Biology → Photosynthesis
  3. Explore in KE, practice in SPL, assess in SBCAT
  4. Note the hierarchical structure and prerequisite chains

Part B: Competency Mode

  1. Switch to "Competency-Based Learning"
  2. Navigate: Data Literacy → Statistical Analysis → Advanced (3/3)
  3. Explore in KE, practice in SPL, assess in SBCAT
  4. Note the proficiency-based structure and flexible pathways

Comparison Questions:

  • How is navigation different between modes?
  • How is content structured differently?
  • Are the learning modes (KE/SPL/SBCAT) identical? (Yes!)
  • Which philosophy matches your institutional approach?
Try Curriculum Mode Try Competency Mode

6.3 Group Discussion: Philosophy Selection (5 minutes)

  • Which philosophy aligns with your institution's approach?
  • Could you use both philosophies for different subjects?
  • How does philosophy-agnostic architecture benefit learners who transfer?

Part 7: xAPI Analytics & Learner Models (30 minutes)

1:00 - 1:30 PM

7.1 xAPI Learning Analytics Overview (10 minutes)

Key Concepts:

  • xAPI (Experience API) for distributed learning records
  • Cookieless authentication: xAPI LRS as session storage
  • Statement structure: Actor-Verb-Object-Result-Context
  • Learner model persistence and incremental updates
  • Cross-institutional learner portability
Key Innovation: "xAPI LRS serves triple duty: authentication, analytics, and learner model storage"

7.2 Demo: Learner Model & Analytics Dashboard (10 minutes)

Part A: Learner Model

  1. Navigate to "My Learning Profile"
  2. Show comprehensive learner model JSON:
    • Knowledge state (mastery scores per concept)
    • Cognitive characteristics (problem-solving, metacognition)
    • Affective state (confidence, motivation, emotion)
    • Learning preferences (visual, auditory, kinesthetic)
    • Learning history (systems used, time spent, sessions)

Part B: Analytics Dashboard

  1. Show xAPI statement stream (all learning activities)
  2. Filter by verb: "experienced" (KE), "attempted" (SPL), "answered" (SBCAT)
  3. Visualize learning trajectory over time
  4. Show mastery progression for photosynthesis concepts
View Teacher Dashboard

7.3 Hands-On Activity: Explore Your Learning Data (10 minutes)

Instructions:

  1. Access your learner profile
  2. Review your knowledge state (concepts mastered, gaps)
  3. View your learning history (time in each mode)
  4. Check your affective state (confidence level)
  5. Download your xAPI statements (JSON export)

Observation Prompts:

  • What insights does the learner model provide?
  • How accurate is the mastery estimation?
  • What would you want to see added to the analytics?

Part 8: Implementation & Integration (30 minutes)

1:30 - 2:00 PM

8.1 Use Cases & Deployment Scenarios (10 minutes)

Use Case 1: Classroom Supplementation

  • Teachers assign KE exploration for homework
  • In-class SPL problem-solving with teacher facilitation
  • SBCAT for formative assessment (exit tickets)

Use Case 2: Fully Online Course

  • Self-paced learning with AI recommendations
  • Adaptive sequencing (KE → SPL → SBCAT loops)
  • Instructor dashboard for monitoring progress

Use Case 3: Corporate Training

  • Competency-based onboarding and upskilling
  • Scenario-based SPL for job-relevant skills
  • SBCAT for certification and credentialing

Use Case 4: Remediation & Intervention

  • SBCAT pre-assessment identifies gaps
  • Targeted KE exploration for gap-filling
  • SPL practice with scaffolding for struggling learners

8.2 Technical Integration Overview (10 minutes)

Deployment Options:

  • Cloud-hosted: Google Cloud Run (stateless, autoscaling)
  • Self-hosted: Docker container on institutional servers
  • Hybrid: Cloud for LLM gateway, local for data storage

Integration Points:

  • LMS Integration: LTI 1.3 for Canvas, Moodle, Blackboard
  • SSO: OAuth 2.0 (Google, Microsoft, SAML)
  • LRS: xAPI compatible (Learning Locker, Veracity, Yet Analytics)
  • API: REST API for custom integrations

Data Privacy & Security:

  • Cookieless (no tracking, GDPR compliant)
  • Data sovereignty (customer's own LRS)
  • JWT authentication (stateless, secure)
  • Content Security Policy (XSS protection)

8.3 Group Discussion: Implementation Planning (10 minutes)

Breakout Groups (3-4 people):

  1. What use case best fits your institution?
  2. What technical barriers might you face?
  3. What stakeholders need to be involved?
  4. What would success look like in 6 months? 1 year?

Report Back: Each group shares one key insight

☕ BREAK (10 minutes) - 2:00 to 2:10 PM

Part 9: Advanced Features & Future Directions (20 minutes)

2:10 - 2:30 PM

9.1 Advanced Agent Features (10 minutes)

Current (30 production-ready agents):

  • Core learner modeling, tutoring interaction, content generation
  • Pedagogical strategies (Socratic, EMT, cognitive apprenticeship)
  • Assessment and analytics

Coming Soon (70 extended agents):

  • Neuroscience-informed learning: Neuroplasticity, working memory, sleep consolidation
  • AI ethics: Algorithmic fairness, privacy protection, bias detection
  • Domain specialists: STEM, literacy, arts, critical thinking
  • Longitudinal learning: Career pathways, lifelong portfolios
  • Real-world integration: Project-based learning, workplace simulation
  • Parent & community engagement: Family involvement, stakeholder communication

Implementation Roadmap: 4-year phased deployment (Phase 1-4)

9.2 Research Opportunities (10 minutes)

Research Questions:

  1. Optimal mode sequencing: Is KE → SPL → SBCAT always best?
  2. Agent effectiveness: Which agents have the most impact on learning?
  3. Cross-institutional portability: Can learner models transfer between schools?
  4. Ethical AI: How do fairness and bias agents improve equity?
  5. Neuroscience validation: Do multimodal approaches enhance memory?

Collaboration Opportunities:

  • Educational data mining (agent decision patterns)
  • Learning analytics (longitudinal studies)
  • Cognitive science (neuroscience of adaptive learning)
  • AI ethics (fairness, bias, transparency)

Part 10: Wrap-Up & Next Steps (30 minutes)

2:30 - 3:00 PM

10.1 Q&A Session (15 minutes)

Open floor for questions

Common Questions (Be Prepared):

  • "What if students game the system?"
    → Agent monitoring detects gaming patterns; system adapts
  • "How much does LLM content generation cost?"
    → 94-96% token reduction via caching; <$0.01/student/hour
  • "What if the LLM generates incorrect content?"
    → Validation layers, fallback content, educator review dashboard
  • "Can I customize content for my curriculum?"
    → Yes, custom prompts and domain model import
  • "How is this different from Khan Academy/Duolingo?"
    → 100-agent ITS, multi-philosophy, explainable AI, three modes

10.2 Resources & Next Steps (15 minutes)

Takeaway Materials:

  • Workshop slides (PDF)
  • UALS documentation links
  • Sample learner data (anonymized)
  • Implementation checklist
  • Contact information for support

Next Steps:

  1. Pilot Program: Apply for 3-month pilot at your institution
  2. Office Hours: Monthly Q&A sessions for implementers
  3. Community Forum: Join UALS educator community (Discord/Slack)
  4. Follow-Up Workshop: Advanced features (agents, analytics, customization)

Call to Action:

"Ready to transform adaptive learning at your institution? Let's schedule a follow-up call to discuss your specific needs."

Contact:

  • Email: pilot@uals.edu
  • Office hours: First Friday/month, 10 AM PT
  • Community: discord.gg/uals-educators

Workshop Agenda Summary

UALS WORKSHOP AGENDA - 3 HOURS MORNING SESSION (9:00 AM - 12:00 PM) 9:00 - 9:30 Introduction & System Overview 9:30 - 10:00 Knowledge Explorer (KE) + Hands-On 10:00 - 10:40 Socratic Playground (SPL) + Hands-On 10:40 - 10:50 BREAK 10:50 - 11:20 Scenario-Based CAT (SBCAT) + Hands-On 11:20 - 12:00 100-Agent Architecture + "Show AI Thinking" Demo 12:00 - 12:30 LUNCH BREAK AFTERNOON SESSION (12:30 PM - 3:00 PM) 12:30 - 1:00 Multi-Philosophy Framework + Hands-On 1:00 - 1:30 xAPI Analytics & Learner Models 1:30 - 2:00 Implementation & Use Cases 2:00 - 2:10 BREAK 2:10 - 2:30 Advanced Features & Research 2:30 - 3:00 Q&A + Wrap-Up + Next Steps HANDS-ON ACTIVITIES: ✓ Explore photosynthesis in Knowledge Explorer ✓ Solve problems in Socratic Playground ✓ Take adaptive assessment in SBCAT ✓ View "Show AI Thinking" workflow visualization ✓ Compare curriculum vs. competency-based learning ✓ Explore your learner model and analytics TAKEAWAYS: ✓ Access to demo system for 30 days ✓ Workshop slides and documentation ✓ Implementation resources ✓ Follow-up consultation (optional)