Class Review Demo - The Human Code

Session A - AI Mechanics

Live AI Demo: Prompting Personas

★★★★★

5 / 5 - Landmark session. Set the tone for the entire course.

What Went Well

Students drove every prompt themselves. Each student named a persona - Harvard professor, empathetic teacher, substitute who doesn't care - and owned the output. 100% participation without enforcement.
The contrast between outputs was the single clearest demonstration of how AI works. Students could see it mirroring their instructions in real time rather than thinking independently.
Student 3's "why should we try if there's no accountability?" was a genuine philosophical question - generated spontaneously from the exercise with no setup from the teacher.
Student 1 noticed mid-exercise that changing three words in a prompt shifted the entire output. That level of analytical observation is exceptional for this age group.
The absurdist persona (substitute who gives everyone 100%) created group energy that carried into the follow-up discussion. The best engagement often comes from students' own humor.

Opportunities

Students watched and reacted; they didn't write their own prompts independently. A follow-on session where each student drafts a prompt to the same task would deepen retention significantly.
The richest philosophical moment - "who's responsible for bad AI output?" - surfaced at the end and got cut off. Build 10 minutes of protected reflection time into future high-engagement sessions.
Students 5 and 6 were primarily observers. Assigning each student a required "prompt direction" before the session ensures every voice shapes the exercise.

Student Snapshot This Session

Student 1

Drove the Harvard persona. Spotted output changes in real time. Strongest analytical engagement of the session.

Student 2

Led the empathetic teacher prompt with real ownership. One of her best sessions - personal voice came through clearly.

Student 3

Substitute teacher idea was the highlight - funny, creative, and sparked more discussion than any planned prompt.

Student 4

Raised the accountability question unprompted. A genuine philosophical insight. Worth returning to explicitly.

Student 5

Present and listening. Minimal output this session - structured role assignment would help next time.

Student 6

Engaged with the outputs but didn't drive a prompt. Check in privately - may need a lower-stakes entry point.

Adapt the Next Lesson

Give Students the Keys

This session proved what happens when students direct the exercise. Next session, give each student an identical task and ask them to write their own prompt for it independently. The output comparison - six different prompts, six different results, same task - teaches more about how AI works than any explanation.

Open the next class by returning to Student 4's accountability question: "Last time someone asked - if AI just does what we tell it, who's responsible when it goes wrong? I've been thinking about that all week." That framing elevates a student's instinct and sets up a richer discussion than cold-starting from theory.

Assign Student 5 and Student 6 a specific persona direction before the next session. "Your job next class is to give us a prompt for a first-year teacher who's terrified of technology." Pre-assignment eliminates the pressure of thinking on the spot.

How this was generated: This review is synthesized from the session transcript using pattern analysis across student contributions, question quality, and pedagogical moments. Student identities and voice signatures are anonymized before any analysis is run. This is illustrative demo data.

Session B - Theological Framework

Guest Expert: Theology of Work & AI

★★★★☆

4 / 5 - Strong framework. Student role was too passive.

What Went Well

"AI rearranges rather than creates" became the North Star concept for the entire course. Memorable, theologically grounded, and immediately applicable to every AI example that followed.
The guest disclosed that AI had helped prepare his presentation - that moment of honesty created immediate credibility and mirrored the transparency modeled throughout the course.
Student 1's question connecting AI literacy to intergenerational advantage was the best student-generated insight of the session. It showed synthesis, not just absorption.
The Yeti cup analogy - "creation by cooperation" - is the kind of concrete, funny anchor that students still reference three months later.

Opportunities

Students shifted to observer mode during the guest session. A pre-assigned "question team" - each student responsible for one prepared question - would have maintained their active role.
The conversation stayed at an educator level for long stretches. A deliberate pivot - "does what our guest said match your experience in other classes?" - would have re-centered students.
Students 5 and 6 had minimal transcript presence. Guest sessions need a more structured Q&A format to ensure every student contributes at least once.
The hallucination concept surfaced briefly but deserved its own session. It's too important to leave as a footnote.

Adapt the Next Lesson

Apply "Rearranges Rather Than Creates" Immediately

The framework needs an application session before it fades. Open the next class with a 10-minute exercise: "give AI a creative task - write a song, design a logo, come up with an invention idea. Then ask: did it create this, or rearrange things it already knows?" Students will discover the answer themselves rather than being told it.

The hallucination mention deserves its own lesson. Frame it as: "our guest said AI can lie without knowing it's lying. What does that mean for how we trust it?" That's one of the most important questions in the course and it got 90 seconds. Give it 45 minutes.

Consider having the guest return for final presentations. If he's already anchored the theological framework, having him in the room when students present their projects closes a loop they'll remember.

How this was generated: Synthesized from the session transcript. Student identities anonymized. Illustrative demo data.

Session C - Critical Thinking

AI Hallucinations & When to Trust It

★★★☆☆

3 / 5 - Right topic, needed a stronger hook and structure.

What Went Well

Students' existing skepticism about AI ("it gets things wrong") gave the session a ready foundation - they weren't starting from zero.
The live demonstration of an AI confidently stating something wrong landed better than any explanation would have. Seeing it happen is different from being told it happens.
Student 2 asked how you know when to trust AI versus when to verify - exactly the right question. It deserved more airtime than it got.

Opportunities

The session needed a stronger hook. Starting with a famous real-world AI error - a lawyer citing fake cases, a medical chatbot giving dangerous advice - would have grounded the concept in stakes before the demonstration.
Discussion stayed abstract. Students said hallucinations are a problem but couldn't articulate when they'd personally been misled. "Has AI ever told you something wrong?" as an opener would have made it immediate.
No takeaway framework. Students left knowing AI can be wrong but not with a protocol for deciding when to verify. The next session should give them a simple rule to apply.

Adapt the Next Lesson

Give Students a Verification Protocol

Students now know AI can be wrong. The gap is knowing what to do about it. Next session, give them a three-question test: Is this claim verifiable? Is it consequential if wrong? Would I act on it without checking? If yes to all three, verify. That's a usable rule they can apply to their own projects.

Apply the protocol to their final projects directly. "Where in your app could AI give bad advice? What happens if it does? What's your safeguard?" This connects the abstract lesson to their own work and sets up the Human Dignity pillar conversation they'll need for project presentations.

Consider showing the MIT deepfake video before the next discussion. Hallucination and deepfakes are the same underlying problem - AI producing confident misinformation - and the visual impact of seeing a convincing fake outperforms any lecture.

How this was generated: Synthesized from the session transcript. Student identities anonymized. Illustrative demo data.

Session D - Final Project Phase

Project Critique Workshop

★★★★☆

4 / 5 - Format is working. Structure needs tightening before presentations.

What Went Well

Peer critique was substantive, not performative. Students pushed on scope, feasibility, and user definition in ways that genuinely strengthened the projects under discussion.
The "constraints make products better" insight surfaced organically - multiple students applied it across different projects without prompting. That's curriculum sticking.
Student 4 received challenging feedback on her project's technical scope and immediately adapted the concept in real time. That's the skill this course is building.
Student 1's cross-project observations - connecting lessons from one student's scope decision to another's - showed the class thinking as a cohort, not just as six individuals.

Opportunities

Class started 5+ minutes late - a pattern worth interrupting before final presentations. A student timekeeper role would address this without requiring the teacher to be the enforcer.
Student 1 dominated the first two project discussions. Quality contributions, but other students need more direct on-ramps. Try assigning the "first question" role to quieter students.
Student 6 gave surface-level feedback ("it's good, it'll inspire people") without follow-through. A targeted pre-session question - "what's one thing you don't understand about this project?" - would have drawn out something more substantive.
Only two of six projects were reviewed in the visible session window. A tighter time format ensures all projects get equitable airtime.

Student Snapshot This Session

Student 1

Sharpest product instincts in the room. Needs an occasional reminder to let others develop their thinking before he steps in.

Student 2

Raised a market segmentation question that reshaped the discussion. Brief but high-value contribution.

Student 3

Offered a solid constraint question but didn't develop it further when pushed. More airtime would help.

Student 4

Received feedback well and built on it in the same session. Real-time iteration is a sophisticated skill - acknowledge it.

Student 5

Best analytical question of the session - a constraint problem others hadn't spotted. Growing into his role as a critic.

Student 6

Surface-level feedback. Pre-session prep question recommended. Not seeing best thinking in this format.

Adapt the Next Lesson

Tighten the Format Before Final Presentations

The peer critique format is working - students are doing real product critique, not cheerleading. Two things need to tighten before presentations: time management and participation equity.

Assign a student timekeeper. Give them a phone timer and the job of calling "two minutes" and "wrap up." This moves clock-watching off the teacher and gives a student a visible role that isn't about their own project.

For Student 6: before next class, give him one question to sit with - "what's one thing about your project that still doesn't quite make sense to you?" If he thinks about it before class rather than on the spot, the answer will be real instead of reflexive.

Assign Student 5 the "first question" role for the next project review. He's capable of sharp critique - he just needs an explicit on-ramp that doesn't feel like performance.

How this was generated: Synthesized from the session transcript. Student identities anonymized. Illustrative demo data.