Presenter notes outline · Morning
From Prompts to Possibilities: AI-Assisted Learning in the K–12 Classroom
Timothy M. Stafford, PhD · Full Professor; Program Director; Director of Graduate Online Academic Affairs; Lead Researcher and Creator of the RU Institute AI Learning Lab · St. Thomas University · 10:30–11:30 a.m. Eastern · Room 103BC
Run of show
| 10:30–10:33 | Opening | 3:00 |
| 10:33–10:42 | Theory and design foundations | 9:00 |
| 10:42–10:51 | Classroom examples | 9:00 |
| 10:51–11:03 | WALK™ practice and debrief | 12:00 |
| 11:03–11:08 | Current research: CourseSpark™ | 5:00 |
| 11:08–11:15 | Three-year vision and next step | 7:00 |
| 11:15–11:30 | Q&A (15 minutes) | 15:00 |
1. Title
0:30 · 10:30Opening
I am Tim Stafford from St. Thomas University in Miami Gardens.
- I serve as Full Professor, Program Director, and Director of Graduate Online Academic Affairs.
- I am also Lead Researcher and Creator of the RU Institute AI Learning Lab.
- Confirm the clicker works with PageDown before people arrive.
- Have the printed WALK design canvas on tables or ready to hand out.
Transition: Before I tell you anything about AI, I want to tell you why I care about this.
2. When answers are everywhere, what will our students still need to be able to do themselves?
1:00 · 10:30:30Opening
When answers are everywhere, what will our students still need to be able to do themselves?
- This is a human question, not just a technology question: what remains the student's responsibility when tools can produce answers?
- A student can get a polished paragraph in seconds and still need to decide whether it is true, whether it fits, and whether they believe it.
Transition: Turn to someone near you. You will have two minutes.
3. Turn and talk: one capability you would protect
1:00 · 10:31:30Opening
Name one thing students should still do themselves, even with AI nearby.
- Capabilities worth protecting include independent thinking, asking good questions, distinguishing true from false, and explaining reasoning.
- Persistence, collaboration, judgment, and integrity matter beyond producing a finished product.
- Take one minute with a neighbor, thirty seconds each, to name a capability and explain why; if sitting alone, jot it on the canvas.
- At time, take two quick shares from different grade bands, one sentence each.
- Note the words you hear. You will return to them in the vision segment.
- Think for themselves, ask good questions, tell true from false, explain their reasoning, persist, collaborate. Administrators often say judgment or integrity.
Transition: Almost nothing you named was about producing a finished product. Hold on to that.
4. Our hour together
0:30 · 10:32:30Opening
We will move from theory to design choices, classroom examples, your own redesign, and current research work.
- The first forty-five minutes combine presentation and practice, followed by fifteen minutes for questions.
- The sequence connects theoretical ideas to classroom design before moving into your own redesign.
Transition: Start with the sentence I most want you to remember.
5. AI should expand what students can think through, create, and explain.
0:30 · 10:33Theory and design foundations
AI should expand what students can think through, create, and explain.
- Think through: AI can offer counterexamples and alternative views for students to test, widening the problems they can reason about.
- Create: AI can lower barriers to more ambitious work, while the decisions remain with students.
- Explain: the student's own explanation is the final evidence of learning. If AI makes that explanation weaker, the design needs work.
Transition: So what does thinking actually look like in a classroom?
6. Central theory: pedagogy and heutagogy as partners
2:30 · 10:33:30Theory and design foundations
Teachers keep teaching while planning for more learner choice, reflection, and self-direction over time.
- Pedagogy retains modeling, explaining, and structuring. Greater learner agency does not mean teachers stop teaching.
- Heutagogy means self-determined learning: learners increasingly decide what to learn, how to learn, and how they will know they learned it.
- Agency, capability, reflection, and learning how to learn sit at its heart.
- Heutagogy grew from adult learning. In K–12, it partners with pedagogy rather than becoming a stage children graduate into.
- Every AI-assisted task should leave a real decision with the student, within teacher-designed structure, boundaries, and support.
- Keep theory brief and practical. Citations are in the notes outline.
Transition: Three supporting ideas explain how that agency actually grows. First: talk and support.
7. Thinking grows through talk and fading support
1:15 · 10:36Theory and design foundations
Thinking grows through talk and fading support.
- Social constructivism emphasizes understanding built through shared activity and conversation, then taken on by the individual.
- Scaffolding is temporary support that decreases as learners become capable. Every scaffold needs a plan for fading.
- Dialogic teaching uses questions and answers that build on one another, rather than simply checking recall.
- AI and teachers can ask questions back instead of handing over answers. Students explain ideas to a peer before accepting them.
- In the seed class, the teacher turns an AI suggestion into a question the class tests together.
Transition: Second idea: students thinking about their own thinking.
8. Students can learn to watch and steer their own thinking
1:00 · 10:37:15Theory and design foundations
Students can learn to watch and steer their own thinking.
- Metacognition involves knowing about and monitoring your own thinking: what you know, what you doubt, and what to do next.
- Self-regulated learners plan, monitor, and reflect. These habits can develop and support learning how to learn.
- Useful prompts include: How sure am I? What would change my mind? What changed in my thinking?
- After finding a flaw in an AI explanation, students write what they now think and why.
- Build brief reflection and confidence checks into every AI-assisted task.
Transition: Third idea: does learning travel?
9. Learning should travel to new situations
1:00 · 10:38:15Theory and design foundations
Using an idea somewhere new is the real test of learning, and it rarely happens on its own.
- Perkins and Salomon's work on transfer reminds us that applying learning in new contexts needs deliberate teaching.
- Name the principle, vary the examples, then check a new case without AI or scaffolds.
- After explaining a disappearing puddle, students independently explain a sweating cold glass.
- Performance with the tool present is not yet evidence that students can reason without it.
Transition: How much support, for whom? That depends on readiness.
10. Scaffold agency by readiness
1:00 · 10:39:15Theory and design foundations
Scaffold agency by readiness, not grade level alone.
- In elementary settings, teachers run the AI interaction and model checking; children ask questions, predict, and compare with observations.
- In middle school, teachers set boundaries and evidence requirements; students compare AI claims with data and explain what they trust.
- In high school, teachers design open problems and coach judgment; students choose approaches, challenge outputs, and justify decisions.
- Release decisions gradually as readiness grows, rather than assuming every learner in a grade needs the same support.
- Direct student use requires age-appropriate, institution-approved tools within school guidance. Teacher-mediated activities can work without student accounts.
- Questions about terms of service and age limits. Answer: follow your district's approved tool list and guidance; this design works even with no student accounts at all.
Transition: To keep our eye on what we are actually building in students, I use a simple lens.
11. From theory to classroom to current research
1:00 · 10:40:15Theory and design foundations
Theory connects classroom design choices to current research questions, not settled findings.
- Pedagogy and heutagogy leave real decisions with students, such as choosing a seed idea to test. Research asks whether dialogue supports agency and judgment.
- Dialogue and scaffolding turn AI output into class questions and plan for fading support. Research asks how much teacher support is needed, and when.
- Metacognition uses reflection and confidence prompts, such as “What changed in my thinking?” Research asks whether reconsider prompts make thinking visible.
- Transfer requires an independent new case, such as explaining a sweating glass. Research asks whether reasoning carries over without the tool.
Transition: One more distinction: established theory versus my developing framework.
12. Human Capacity/Capability Framework™
0:45 · 10:41:15Theory and design foundations
The Human Capacity/Capability Framework™ is a developing design lens, not an established theory or validated measure.
- Human capability means what a person can do with what they know: reason, judge, adapt, and keep learning.
- Visible reasoning means students show how they got there; better judgment means they weigh claims and choose well.
- Learning transfer means using knowledge somewhere new; learning to become means growing as learners and people.
- These four outcomes bring established research traditions into language teachers can use for design conversations.
- AI sits at the center as a possible catalyst, never the goal.
Transition: Before examples, a few practical guardrails.
13. Five habits that keep it safe and fair
0:45 · 10:42Classroom examples
Five practical habits keep AI use safe and fair, with district policies coming first.
- Mediate for young learners: teachers run the tool while children do the thinking.
- Use age-appropriate, institution-approved tools within school guidance.
- Keep data private: use fictional data, not student names or personal details, in prompts.
- Check AI claims against trusted sources. Confident wording does not guarantee accuracy.
- Offer printed outputs so students can participate without accounts or devices. That is an equity design choice, not a compromise.
Transition: With that foundation, let us look at the aims we are designing for.
14. Give AI a job. Keep the thinking with students.
0:45 · 10:42:45Classroom examples
Give AI a job. Keep the thinking with students.
- A question coach asks follow-up questions rather than supplying answers; students still have to explain their thinking.
- An example maker produces samples, including flawed ones, for students to evaluate.
- A sparring partner offers a counterargument that students must answer.
- A feedback mirror suggests revisions that students accept or reject with reasons.
- Avoid making AI the answer machine for core learning goals. That removes the thinking we need to see.
- Ask the room: which of these roles would be safest to start with in your setting? Expect example maker and question coach.
Transition: Now to the bigger idea underneath all of this: how teaching and learner agency work together.
15. A finished product is not the same as visible reasoning
1:00 · 10:43:30Classroom examples
A finished product is not the same as visible reasoning.
- A polished essay, poster, or correct answer shows what was submitted, but not necessarily what the student learned.
- Visible reasoning includes students' choices, their reasons, and what they checked and changed.
- Assessing only the artifact lets shortcuts stand in for learning. The response is better evidence design, not assumptions about student character.
- Integrity still matters. Asking for reasoning gives students a way to show their own work and discuss AI use openly.
- Someone may raise cheating. Respond: integrity still matters, and visible reasoning is one of the strongest supports for it, because it gives students a reason and a way to show their own work.
Transition: Which leads to a sentence I want to say out loud.
16. Why did some seeds grow faster?
0:45 · 10:44:30Classroom examples
Children test AI explanations against their own seedling chart, because their observations are the evidence.
- In this fictional second- or third-grade example, children measure bean seedlings for two weeks and notice that some cups grow faster.
- The teacher asks AI for possible explanations in front of the class and reads the response aloud.
- Children compare those explanations with their own chart rather than treating the AI response as the answer.
- The teacher operates the tool; children do the checking. AI is one more voice to evaluate.
Transition: Here is what that interaction might sound like.
17. Children check the ideas
1:30 · 10:45:15Classroom examples
AI offered ideas; the class decided what to test.
- In this fictional example, the teacher turns AI suggestions into questions young children can investigate, without student devices.
- Children compare same water, same cup, and different light, then propose moving two shelf cups to the window.
- Choosing a test, predicting, and using their own data makes curiosity, critical thinking, and agency visible.
- AI is an optional outside voice to evaluate, not a requirement for this kind of learning.
- Elementary teachers often say: I already do this without AI. Affirm that strongly. AI adds a quick, outside voice to evaluate. It is optional, not required.
Transition: Now to middle school, where students begin to handle AI claims and data more directly.
18. Does the general claim fit our town?
1:30 · 10:46:45Classroom examples
A general claim can be reasonable and still not fit a specific context.
- Both the printed AI summary and the town data are fictional samples created for this session.
- The summary downplays outdoor water use, while the town data shows outdoor use is about half in summer and drops sharply in winter.
- Students identify which claim fits, which needs a qualifier, and what evidence convinced them.
- The important distinction is season and place, not simply whether a claim sounds plausible.
- Teachers may ask where to find real local data. Suggest your local water utility's public reports, checked by the teacher first.
Transition: And in high school, the same habit applies to interpreting sources.
19. Students push back on an overreach
2:45 · 10:48:15Classroom examples
Students source, corroborate, and revise an overgeneralization into a narrower, better-supported claim.
- The mill owner's letter and worker's diary are fictional sources describing the same town's flood.
- The AI interpretation smooths over disagreement by claiming everyone recovered quickly and worked together.
- Delayed wages challenge that claim, while the owner's perspective gives students reason to question the positive account.
- Students conclude that recovery was uneven: AI offers an interpretation, and students decide what it is worth.
Transition: So how do you design that on purpose? Four steps called WALK.
20. Four steps to redesign one task
0:30 · 10:51WALK™ practice and debrief
WALK connects the thinking that matters to student decisions, a small test, and aligned evidence.
- W: What thinking matters? Name the thinking students need to do.
- A: Ask who does what. Keep decisions with students and set the scaffold.
- L: Launch a small test. Keep the redesign safe and manageable.
- K: Keep evidence aligned. Include reflection and an independent transfer check.
Transition: Here is one task, before and after.
21. Same topic. Different thinking.
1:00 · 10:51:30WALK™ practice and debrief
Same topic. Different thinking.
- A finished water-cycle report can hide who did the thinking because AI could write the whole thing.
- The redesign asks students to explain why a puddle disappears by finding and fixing a flaw in one of two AI explanations.
- The small test is one 20 minute segment with one class, with no student devices required.
- Evidence includes the flaw, correction, and reflection, followed by an independent explanation of a sweating glass.
- The task is harder to shortcut because students must exercise their own judgment and explain their reasoning.
Transition: Your turn. Seven focused minutes.
22. Redesign one of your own tasks with WALK
6:30 · 10:52:30WALK™ practice and debrief
A WALK redesign names the human thinking, limits AI's contribution, and checks independent transfer.
- Name thinking with an observable verb, not just a topic: explain why a plant in a closet turns pale, rather than photosynthesis.
- AI can contribute a printed flawed example, while students retain the decision about what needs correcting and why.
- Student reasoning and a short reflection make the thinking visible; a transfer task checks what students can do without AI.
- Use seven minutes to redesign one real upcoming task, aiming for one clear line per box.
- Start a visible 5 minute timer. Show the administrator slide for about 30 seconds within this block, then return.
- Circulate and push gently for an observable verb in the W box.
- At 4 minutes: if you are stuck on the AI role, use example maker with a printed flawed sample.
- At 6 minutes: make sure your transfer task has no AI support.
- Common struggle: naming a topic (photosynthesis) instead of thinking (explain why a plant in a closet turns pale).
- Some choose AI as an answer source. Ask: what decision does the student make with it?
Transition: Administrators, here is your parallel path.
23. WALK a support or observation process
0:30 · 10:59WALK™ practice and debrief
Observe student reasoning rather than counting how often a tool is used.
- WALK applies to a coaching cycle, walkthrough form, or PLC protocol, not only classroom tasks.
- W names what students should be doing; A clarifies the roles of the coach, teacher, team, and tools.
- L keeps the test focused: one team, one cycle, one look-for.
- K prioritizes reasoning samples over usage counts, which tell us about access rather than learning.
Transition: Return to the activity slide and keep the timer visible.
24. Pair check: is the thinking really visible?
2:45 · 10:59:30WALK™ practice and debrief
Aligned evidence shows what students actually do with the thinking that matters.
- The central check is concrete: what will we see or hear students do?
- A poster alone may not show the explanation that a task was meant to develop.
- A partner can catch a mismatch between the intended thinking and the evidence being collected.
- Use three minutes to explain, question, switch roles, and revise one line on the canvas.
- Call the switch at 50 seconds. Leave 20 seconds to revise.
- Partners often catch evidence that does not match the thinking, such as collecting a poster when the goal was explanation.
Transition: Quick debrief.
25. Name evidence you could actually observe
0:45 · 11:02:15WALK™ practice and debrief
Name evidence you could actually observe.
- “Students understand” names a goal, but not what a student would do to demonstrate it.
- Explaining why they rejected a suggestion makes student judgment observable.
- Finding and fixing a flawed step shows more than simply engaging with AI.
- Solving a new case alone provides observable evidence of transfer.
- Write your strongest observable evidence statement in the K box.
Transition: Now to our current work in this space.
26. What we want to learn
1:00 · 11:03Current research: CourseSpark™
These are research aims and questions, not results.
- Visible reasoning: does AI-assisted dialogue help students show their thinking?
- Agency and judgment: do students make and defend more of their own decisions?
- Transfer: can students apply their learning later, without the tool?
- Teacher support: how much guidance is needed, and when? Heavy support needs would be useful to understand, not a failure.
- Do not mention participant numbers, findings, or pilot details. Keep pilot details out until they are verified.
Transition: One illustrative application we are thinking with is called CourseSpark.
27. How the design tries to put these ideas to work
1:15 · 11:04Current research: CourseSpark™
CourseSpark illustrates design intent, not proof that these approaches work.
- Socratic dialogue draws on dialogic learning: the tool asks questions rather than gives answers.
- Justify prompts ask “Why do you think that?” to make reasoning visible.
- Reconsider prompts ask “What might change your mind?” to invite students to check their own thinking.
- Transcripts let teachers review thinking and decide what support to provide next, or fade.
- Reflection asks what changed and remains uncertain, connecting to self-regulation. Whether it supports transfer remains a research question.
Transition: Here is one illustrative walkthrough.
28. One illustrative walkthrough
1:45 · 11:05:15Current research: CourseSpark™
The transcript lets a teacher review how thinking changed, not just the final answer.
- “Where did the water go?” asks the student to explain what happened beyond saying the sun dried the puddle.
- The cloudy, cool day prompt invites reconsideration: heat speeds evaporation but is not the only reason.
- The revised explanation is more careful, giving a teacher something to examine when deciding on support.
- This is a written illustration of the design, not real student data or a live product demonstration.
- CourseSpark appears here in a research context, not a sales pitch. Mentioning it does not imply St. Thomas University ownership or endorsement.
Transition: And to be clear about what we know so far.
29. Questions we are asking, not claims we are making
1:00 · 11:07Current research: CourseSpark™
These are questions we are asking, not claims we are making.
- There are no findings to report today and no claims of proven results.
- This is not a claim of K–12 approval or a live integration in your school.
- The questions remain: what reasoning becomes visible, where are teachers needed most, and what evidence would show transfer?
- To follow the work as it develops, find me at the STU vendor space after the session.
Transition: Let me close with a vision of where this could go in three years.
30. From safe experiments to lasting capability
1:30 · 11:08Three-year vision and next step
From safe experiments to lasting capability is an aspiration, not a forecast.
- Year 1 establishes safe, small experiments and baselines through observations, reasoning samples, and teacher workload notes.
- Year 2 refines and shares supported practice across teams, with students evaluating and justifying more.
- Year 3 looks for sustained capability: students questioning, judging, and transferring learning more independently.
- Observations, reasoning samples, independent work, reflections, and workload notes inform the vision. Alone, they do not prove AI caused change.
Transition: None of that happens without a first step. So what is yours?
31. What we hope each group can do by Year 3
1:30 · 11:09:30Three-year vision and next step
The Year 3 vision is growth for students, teachers, and administrators, not a forecast.
- Students ask better questions, evaluate AI suggestions, justify choices, transfer learning, and work more independently.
- The student goals connect questioning and choosing, explaining, judging and reflecting, and applying learning somewhere new.
- Teachers shift from producing content toward designing responsive challenges and interpreting student thinking, with manageable workload.
- Administrators support teacher capacity, equitable access, and approved, age-appropriate tools.
- Quality is judged through reasoning samples and observations, not usage counts alone.
Transition: And how would we know whether any of this is happening?
32. What we would look at, and what it cannot prove alone
1:15 · 11:11Three-year vision and next step
Evidence can show patterns worth discussing without proving that AI caused a change.
- Classroom observations, student reasoning samples, and independent work provide practical evidence schools can gather without new systems.
- Student reflections and teacher workload notes add perspectives on learning and the support it requires.
- Many things change during a school year, so these sources alone cannot establish causality.
- Year 1 baselines give later evidence something to be compared with.
Transition: Which brings it back to one small step.
33. Before Friday, try one WALK step with one class or one team.
0:45 · 11:12:15Three-year vision and next step
Before Friday, try one WALK step with one class or one team.
- Keep the commitment focused: one WALK step, one class or team, before Friday.
- Telling a colleague makes the action more likely to happen.
- Take 45 seconds to write your step at the bottom of the canvas, then tell a colleague.
Transition: One sentence to carry out of the room.
34. Teachers design the conditions; students increasingly shape the learning.
1:00 · 11:13Three-year vision and next step
Teachers design the conditions; students increasingly shape the learning.
- This is pedagogy and heutagogy working together.
- Teachers retain a design role while students take an increasing role in shaping their learning.
Transition: Your questions.
35. Want ready-to-use routines? Join us this afternoon.
0:30 · 11:14Three-year vision and next step
This afternoon turns task design into four ready-to-use routines that get students thinking.
- Practice four routines: question before answer; compare and justify; find and repair a flaw; discuss, decide, reflect.
- The session uses different examples from this morning, with hands-on practice and no device needed.
- AI-Assisted Learning That Gets Students Thinking runs from 2:00 to 3:00 Eastern in Room 203A.
- Prize giveaways are also part of the afternoon session.
Transition: Now, your questions.
36. Join the RU Institute AI Learning Lab
0:30 · 11:14:30Three-year vision and next step
Keep learning together through the RU Institute AI Learning Lab.
- The Skool community offers a seven-day free trial.
- Take a course, connect with colleagues, and keep building your professional practice.
- Use the trial to see whether the community adds value to your ongoing professional development.
- Leave the code up for about 20 seconds so phones can scan it.
Transition: Now, your questions.
37. Questions and conversation
14:30 · 11:15Q&A (15 minutes)
Making this work in your building means addressing risks and identifying the support teachers need first.
- Academic integrity: make reasoning visible, require reflection, and be explicit about acceptable use.
- Tool safety: follow the district’s approved list. Every activity today also works with printed outputs.
- Heutagogy is a partner to pedagogy in K–12, not a replacement.
- Research aims are not findings. Results are not yet available.
- Use the fifteen-minute discussion to surface building-level questions, risks, and teacher support needs.
- Keep answers to about two minutes so many voices are heard. Invite longer conversations to the STU vendor space.
- If questions slow, use the prompts on screen, or ask the room: what would make this work in your building?
- At 11:29:30, move to the closing slide.
- Academic integrity: design for visible reasoning, require reflection, be explicit about acceptable use.
- Which tools are safe for children: follow the district's approved list; every activity today works with printed outputs.
- Theory questions: point to the notes outline references; heutagogy is a partner to pedagogy in K–12, not a replacement.
- Is there research evidence: point to the research aims; be honest that results are not yet available.
Transition: Thank you.
38. Closing and STU booth invitation
0:30 · 11:29:30Q&A (15 minutes)
Keep the conversation going through follow-up and the afternoon session.
- Thank you for your time and thinking this morning.
- The STU vendor space is a place to continue the conversation; email follow-up is welcome at tstafford@stu.edu.
- The afternoon session in Room 203A offers completely different examples and routines.
Verified references
- Alexander, R. J. (2008). Towards dialogic teaching: Rethinking classroom talk (4th ed.). Dialogos.
- Blaschke, L. M. (2012). Heutagogy and lifelong learning: A review of heutagogical practice and self-determined learning. The International Review of Research in Open and Distributed Learning, 13(1), 56–71. https://doi.org/10.19173/irrodl.v13i1.1076
- Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist, 34(10), 906–911. https://doi.org/10.1037/0003-066X.34.10.906
- Hase, S., & Kenyon, C. (2000). From andragogy to heutagogy. UltiBASE. RMIT University.
- Perkins, D. N., & Salomon, G. (1992). Transfer of learning. In T. Husén & T. N. Postlethwaite (Eds.), International encyclopedia of education (2nd ed.). Pergamon Press.
- Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (M. Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds.). Harvard University Press.
- Wood, D., Bruner, J. S., & Ross, G. (1976). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17(2), 89–100. https://doi.org/10.1111/j.1469-7610.1976.tb00381.x
- Zimmerman, B. J. (2002). Becoming a self-regulated learner: An overview. Theory Into Practice, 41(2), 64–70. https://doi.org/10.1207/s15430421tip4102_2
- U.S. Department of Education, Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations. https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf
Presenter note: CourseSpark™ and the Human Capability Framework™ are described as a research context and a design lens. No findings are reported. All classroom data, sources, and AI responses in the decks are fictional samples.