TFIS · Content Review
Complete course audit for the Course: Navigating Ethical Challenges of AI in Student Learning — a 7-session, 14-contact-hour programme for Malaysian HE educators, returning 1–3 September 2026 (rescheduled from 20–22 July). This page is the single source of truth for content review.
Project Brief
01 · Syllabus
The syllabus (full document · HTML) specifies 6 course-level learning outcomes and 8 binding design commitments that govern all downstream deliverables.
Describe the current AI capability frontier and its documented effects on student learning using primary evidence (2026 AI Index, PNAS crutch-effect RCT), and map their own discipline's exposure.
Apply the five-principle lens (beneficence, non-maleficence, autonomy, justice, explicability) to real AI-in-learning incidents, grade severity, and assign responsibility across the educator–student–institution triad.
Explain why AI-text detection fails as an evidentiary basis and redesign assessments using process evidence and the three-lane pattern (🔴 Restricted / 🟡 Permitted / 🟢 Required).
Operate AI competently and ethically through the 4D competencies (Delegation, Description, Discernment, Diligence), maintaining verification logs and disclosure norms.
Draft, red-team, and version an institutional AI-use guideline grounded in the Malaysian regulatory context (PDPA 2010, national AI-governance direction, risk-tier thinking).
Quantify their own capability movement on the EAI-CMM (baseline S1 → retake S7) and commit to a falsifiable 90-day plan with owner, dates, and a kill criterion.
02-slides/s1.html … s7.html02 · Architecture
The course is one argument in seven movements: understand the landscape → build ethical fluency → ship institutional policy. Each phase has a distinct pedagogical mode and produces specific artifacts that feed into the capstone.
S1: AI capability frontier, EAI-CMM baseline, Threat/Gift Matrix
S2: Five ethical principles from cases, bias demo, Ethics Triage
S3: Detector bias evidence, Detector on Trial, successor regime
S4: 4D framework, rubric builder, feedback assistant, verification log
S5: Crutch-effect RCT, assessment redesign, guardrailed tutor, policy paragraph
S6: Institutional maturity scan, gap analysis, drafting brief
S7: Drafting sprint, red-team exchange, patch round, EAI-CMM retake, action plan skeleton
(S8 optional: action plan presentations against summative rubric)
The course is built on seven design axioms, stated to participants as they become load-bearing:
From the AI Fluency Framework (Dakan, Feller & Anthropic, CC BY-NC-SA). Four competencies operating as two loops:
Six-level maturity ladder (SSA-CMM pattern), measured at baseline (S1) and retake (S7). Five pillars × 4 items, max 80:
03 · Sessions
Each session has its own single-file HTML deck (s1.html–s7.html) on the tfis-fluency brand system. The facilitator guide · HTML provides the operational run sheet, and the course pack · HTML carries the full scripts and talking points.
All 15 prompts are model-agnostic, using [BRACKETS] for fill-in values. They are documented in the prompt handouts · HTML and the course pack Appendix A · html.
P01 — Cold-open assignment test
P02 — Capability mapper
P03 — Bias probe: reference letters
P04 — Bias probe: cultural default
P05 — Disclosure statement drafter
P06 — Viva question generator
P07 — Rubric builder
P08 — Feedback drafter
P09 — Vulnerability audit
P10 — Guardrailed Socratic tutor
P11 — Course policy paragraph
P12 — Gap analysis partner
P13 — Guideline v0.1 drafter
P14 — Red-team attacker
P15 — Action plan sharpener
04 · Evidence
All sources are restricted to the source register · HTML. The evidentiary core of the course rests on these studies:
| Study | Used In | Claim | Link |
|---|---|---|---|
| 2026 AI Index | S1, S3, S6 | Capability/gap widening; 88% org adoption; 4 in 5 students using AI; only 6% of teachers say policies are clear | Stanford HAI |
| Liang et al., Patterns 2023 | S3 evidentiary core | 61%+ false-positive detector bias vs non-native English writers; one-prompt evasion; cited inside Guideline v0.1 | DOI |
| Bastani et al., PNAS 2025 | S5 keystone | Crutch-effect RCT: −17% unassisted for GPT Base; guardrails engineered harm away; design determines outcome | DOI |
| Wang & Demszky, Tutor CoPilot 2024 | S1 | Real-time AI-assisted tutoring: +4pp overall, +9pp for lowest-rated tutors; AI amplifies human, doesn't replace | arXiv |
| Anthropic Education Report | S1, S2, S3 | ~50% of student AI use is direct answer-seeking; misuse patterns at scale; faculty usage data (74k conversations) | Anthropic |
| Floridi et al., AI4People 2018 | S2 | Five-principles synthesis (beneficence, non-maleficence, autonomy, justice, explicability) from 84 existing frameworks | DOI |
| Wan et al., EMNLP 2023 | S2 (P03 demo) | Systematic gender bias in LLM-generated reference letters (agentic vs communal language) | ACL Anthology |
| Mollick & Mollick, 2023 | S5 | Seven roles for AI in learning: mentor, tutor, coach, student, simulator, teammate, tool | SSRN |
| Kosmyna et al., MIT 2025 | S2, S5 | Cognitive debt preprint (N=54 — cited with limitations); Opperman attention-support work | arXiv |
The course's core design vocabulary for assessments:
| Lane | Rule | What It Protects | Example |
|---|---|---|---|
| AI-Restricted | Foundational skill the student must own unaided | In-class problem set, viva, closed-book segment | |
| AI-Permitted (with disclosure) | Authentic practice — mirrors professional reality | Take-home analysis + AI-use disclosure + process trail | |
| AI-Required | AI fluency itself as a learning outcome | Submit prompt log + critique of AI output + your improvement |
The course's S6–S7 policy work is grounded in Malaysia's proposed AI Governance Bill and existing statutory framework. The key document is the Malaysia grounding note · HTML which crosswalks the Bill's five principles to the course's five ethics principles and maps the Bill's three-tier risk framework onto the three-lane pattern.
The Bill's five AI Principles (Human Dignity, Transparency, Accountability, Safety, Data Stewardship) align almost one-to-one with the Floridi/AI4People lens taught in S2. The key teaching move: the national direction and the classroom ethics converge on the same five ideas.
The Bill's Tier 2 (High Risk) obligations — risk assessment, documentation, traceability, human oversight — are the EAI-CMM Governance pillar and the verification-log habit from S4, at institutional scale.
05 · Sources & Files
All sources are restricted to the source register · HTML. No other sources are cited anywhere in the course. Primary source: Stanford HAI (2026 AI Index). Secondary: Anthropic, Penn/Wharton, MIT, Oxford, Harvard.
| Source | Used In | URL |
|---|---|---|
| 2026 AI Index Report (Stanford HAI) | S1, S3, S6 | hai.stanford.edu/ai-index/2026-ai-index-report |
| AI-Detectors Biased Against Non-Native English Writers | S3 | Stanford HAI News |
| AI's 'Delusional Spirals' (HAI News, Apr 2026) | S2 | Stanford HAI News |
| Liang et al., Patterns 2023 — detector bias | S3 evidentiary core | DOI: 10.1016/j.patter.2023.100779 |
| Wang, Demszky et al., Tutor CoPilot 2024 | S1 | arXiv:2410.03012 |
| Source | Used In | URL |
|---|---|---|
| Bastani et al., PNAS 2025 — crutch-effect RCT | S5 core | DOI: 10.1073/pnas.2422633122 |
| Mollick & Mollick, "Assigning AI" (2023) | S5 | SSRN |
| Kosmyna et al., MIT Media Lab 2025 (preprint) | S2, S5 | arXiv:2503.05667 |
| Floridi et al., AI4People (2018) | S2 | DOI: 10.1007/s11023-018-9482-5 |
| Wan et al., EMNLP Findings 2023 | S2 (P03) | ACL Anthology |
| AI Fluency Framework (CC BY-NC-SA) | S4 spine | aifluencyframework.org |
| Anthropic Education Report (Apr 2025) | S1, S2, S3 | anthropic.com |
| Source | Used In | Notes |
|---|---|---|
| Personal Data Protection Act 2010 (Malaysia) | S2, S4, S6–S7 | Statutory floor for guideline component 6 (data & privacy) |
| NAIO AI Governance Bill fact sheets | S6–S7 | 5 principles, 3-tier risk framework, Central Authority pattern |
[LOCAL] slots | S6–S7 | MQA programme standards, MOHE guidance, institutional senate |
| File | Lines | Description |
|---|---|---|
| AGENTS.md HTML | — | Project documentation |
| PROJECT-BRIEF.md HTML | 64 | Project brief (updated) |
| syllabus.md HTML | 189 | Syllabus v1.0 — the contract |
| s1.html | 350 | Session 1 — The AI Revolution |
| s2.html | 341 | Session 2 — AI Ethics |
| s3.html | 314 | Session 3 — Academic Integrity |
| s4.html | 207 | Session 4 — Workshop Part 1 |
| s5.html | 250 | Session 5 — Workshop Part 2 |
| s6.html | 241 | Session 6 — Guidelines Pt 1 |
| s7.html | 229 | Session 7 — Guidelines Pt 2 |
| facilitator-guide.md HTML | 110 | Delivery-day run sheets |
| prompt-handouts.md HTML | 69 | P01–P15 handouts |
| templates.md HTML | 142 | D1–D6 + S8 rubric |
| eai-cmm.md HTML | 92 | EAI-CMM (D7) |
| capstone-brief.md HTML | 42 | Capstone one-pager |
| course-pack.md HTML | 739 | Full course pack v1.0 |
| source-register.md HTML | 52 | Restricted source register |
| ofqual-2020.md HTML | 31 | Case: UK Ofqual 2020 |
| st-georges-1980s.md HTML | 21 | Case: St George's 1980s |
| texas-am-commerce-2023.md HTML | 28 | Case: Texas A&M 2023 |
| AI-governance-bill.md HTML | 339 | NAIO Bill fact sheets |
| malaysia-grounding.md HTML | 52 | Bill ↔ course crosswalk |
| print-pack.html | 473 | A4 print-ready HTML |
| print-pack.pdf | — | Generated PDF |
06 · Assessments
Seven instruments (D1–D7) plus the S8 summative rubric. Full templates in templates.md HTML.
| # | Instrument | When | Type | Feeds |
|---|---|---|---|---|
| D1 | Exit Ticket 3-2-1 | S1 (3-2-1), S5 (one-word) | Formative | S3 opening, S6 gap analysis |
| D2 | Verification Log | S4 | Formative, collected | Staff-diligence norm in guideline |
| D3 | Assessment Redesign Sheet | S5 | Formative, portfolio | Assessment-design guidance; component 3 |
| D4 | Case Memo (≤200 words) | S2→S3 | Formative | S3 opening; integrity-procedure thinking |
| D5 | Guideline v0.1 Skeleton | S7 | Capstone | The build — 7 components + version block |
| D6 | 7/30/90 Action Plan | S7→S8 | Summative | Post-course follow-through |
| D7 | EAI-CMM Score Sheet | S1 baseline, S7 retake | Self-assessment | Delta = the course's measurable promise |
| — | S8 Summative Rubric (5 criteria × 4 bands) | Session 8 (optional) | Summative | Pass ≥12 · Distinction ≥17 |
Do I understand what this technology is and isn't? (Items 1–4)
Do my teaching and assessment designs account for AI? (Items 5–8)
Is my integrity regime built on design or on policing? (Items 9–12)
Can I evaluate AI output, and teach students to? (Items 13–16)
Do I operate within (and contribute to) explicit rules? (Items 17–20)
Every participant ships a draft institutional AI-use guideline with these components, ≤2 pages, versioned, owned, and AI-use disclosed:
Eight scenario cards ranked on a severity ladder (Critical / Serious / Manageable / Trivial), tagged by primary principle + primary owner. Strict ranking, no ties.
Card 1: Student submits fully AI-written essay, undisclosed, in an "AI-restricted" course.
Card 2: Lecturer uses free public AI to grade essays, pasting full student submissions including names and IDs.
Card 3: Student with dyslexia uses AI to restructure their own draft; course rules are silent.
Card 4: Lecturer fails a student because a detector reported "98% AI"; no other evidence.
Card 5: Faculty buys AI tutor licenses for one elite programme only.
Card 6: Student uses AI to generate practice quizzes and study plans, discloses cheerfully.
Card 7: Lecturer publishes AI-generated notes containing a fabricated reference; students cite it onward.
Card 8: Department bans all AI use, no detection or redesign; usage continues, silently.
10 · Audit
Use this checklist to verify readiness before the 1–3 September 2026 delivery. Each item links to the relevant source for quick verification.
🟢 Content complete — all 25 files built, reviewed, and committed. Delivery readiness items above require in-week verification of regulatory context and physical logistics.
08 · External Audit
MIT Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training — published 13 August 2026. Co-chaired by Eric Klopfer and Sam Madden, representing undergraduate and graduate students, faculty from every school, and staff. Five months of meetings, research, and outreach. Full report ↗
Charged in January 2026 by Chancellor Nobles, Provost Chandrakasan, and Faculty Chair Levy to (1) assess current AI use at MIT, (2) identify innovations in teaching and assessment, and (3) propose an AI-use policy. The committee quickly concluded that deeper questions about the structure, meaning, and value of a residential MIT education were at stake.
The report finds generative AI “everywhere already” at MIT, with strongly mixed student feelings and instructor attitudes ranging from enthusiastic exploration to outright refusal. Key concerning effects identified:
| # | Principle | Core Meaning |
|---|---|---|
| 2.1 | Be humble | AI is <4 years old in public use; course corrections inevitable. Proposals offered in humility. |
| 2.2 | Be bold | Uncertainty is no excuse for inaction. Patches and duct tape are insufficient; a bold strategic response is required. |
| 2.3 | Put humanity front and center | Nurture shared humanity above all. Example: replacing UROPs with AI agents trades apprenticeship for efficiency. |
| 2.4 | Lean into learning | AI’s threat to familiar teaching is a “blessing in disguise.” Create a new social contract around productive struggle. |
| 2.5 | Teach with intentionality | Backward design from learning goals to assessments. “AI-aware” does not mean “AI everywhere.” |
| 2.6 | No one size fits all | A poetry seminar and a math proof course need different AI relationships. A uniform rule would fail both. |
| 2.7 | Augmentation not automation | Pro-learner AI: expand what students can think about and solve, not replace the “hard fun” of learning. |
| 2.8 | Think beyond the classroom | Education is a cultural practice. AI governance must support whole-human development, not just content delivery. |
3.1.1 Revisit course goals with backward design · 3.1.2 Ensure durable learning through new assessments (oral exams, portfolios, process evidence paired with in-class conversation) · 3.1.3 Emphasize experiential and project-based learning · 3.1.4 Build structured in-person social learning into every subject · 3.1.5 Preserve and expand out-of-class research and career experiences (UROPs, RAs) · 3.1.6 Reconsider grades and incentives — explore competency-based and mastery-based alternatives, including the UK percentage system · 3.1.7 Expand in-person spaces for labs and in-person evaluation · 3.1.8 Provide AI-use policies with justification — every syllabus must state why AI is restricted/permitted/required · 3.1.9 Exercise caution with AI detectors and online exam platforms — avoid adversarial arms race · 3.1.10 Support responsible experimentation in the curriculum (AI Light / AI Heavy pathways)
3.2.1 Define and communicate the value of residential education — “why it matters for students to go to college” · 3.2.2 Strengthen social connection and personal wellbeing — Tech Free Times, shared rituals, mental health · 3.2.3 Encourage instructor disclosure around their own AI use — model the behaviour; avoid double standards · 3.2.4 Teach effective, responsible, and ethical use of AI — distinct but interrelated capabilities · 3.2.5 Recognize and mitigate negative impacts of AI — training-data objections, corporate power, environmental cost · 3.2.6 Acknowledge AI use in theses and other research work — standard disclosure statement
3.3.1 Establish an ongoing AI and Education committee · 3.3.2 Create school/department-level AI Leads · 3.3.3 Fund AI Fellows and an AI Implementation Team · 3.3.4 Create an AI Pilot Fund for instructors · 3.3.5 Provide ongoing training and instructor support (lunch-and-learn, communities of practice, workshops) · 3.3.6 Develop metrics — track AI use, campus engagement, student satisfaction, post-graduation feedback · 3.3.7 Ensure equitable technology access — model-agnostic platform, avoid $200/month divide · 3.3.8 Protect sensitive data and preserve model choice · 3.3.9 Establish privacy, logging, and auditing policies for institutional AI platforms · 3.3.10 Monitor AI costs and environmental impact — data centre energy/water use
While comprehensive, the MIT report is notably silent on several dimensions the TFIS course covers:
Each gap below identifies something the MIT report addresses that the current TFIS course does not substantively cover. Gaps are ranked by materiality to the course’s Malaysian HE audience (Critical / Important / Supplementary).
| # | Priority | Gap | MIT Section | Recommended Course Response | Status |
|---|---|---|---|---|---|
| G1 | Critical | Residential experience & social fabric erosion. MIT devotes §3.2 to reclaiming the residential experience: declining study groups, empty office hours, loss of shared rituals, the “social contract.” The TFIS course has no session or module on the social dimensions of AI’s impact on campus life. | §3.2, §2.8 | Add a 10-minute segment in S1 (after the threat/gift matrix) on social-fabric effects. Add a prompt (P01b) for teams to map AI’s effect on their own institution’s informal learning spaces. | Done |
| G2 | Critical | Grades & incentive-system reform. MIT asks whether grading itself is the problem and recommends exploring competency-based and mastery-based alternatives. The TFIS course redesigns assessments within existing grading structures but never questions the grading system itself. | §3.1.6 | Add a “What if there were no grades?” discussion point to S5 (keystone redesign sprint). Incorporate the UK percentage-system example as a contrast case. | Done |
| G3 | Critical | Instructor AI disclosure & double-standard risk. MIT insists instructors must disclose their own AI use and warns the double standard (faculty use AI; ban students) is already eroding trust. The TFIS course has student-disclosure norms (P05) but no equivalent instructor-disclosure framework, and no treatment of the double-standard problem. | §3.2.3 | Add an instructor-disclosure component to the Guideline v0.1 skeleton (Component 4b). Add a “mirror test” exercise in S6 where participants audit their own AI use alongside their draft policy. | Done |
| G4 | Important | Institutional infrastructure for AI governance. MIT recommends dedicated AI Leads per school/department, an AI Implementation Team, AI Fellows, and an AI Pilot Fund. The TFIS guideline v0.1 (7 components) has no “roles & infrastructure” component. | §3.3.1–3.3.4 | Add an 8th optional component to the guideline skeleton: “Roles, Infrastructure & Funding.” Include a short case study of MIT’s proposed AI Leads model. | Done |
| G5 | Important | Research-apprenticeship displacement (UROP/RAs). MIT specifically warns that faculty may replace undergraduate research assistants with AI agents, ending the apprenticeship model. The TFIS course has no content on research-learning displacement. | §3.1.5, §2.3 | Add a 5-minute touchpoint in S2’s ethics-triage discussion: new scenario card on “Faculty replaces UROP student with AI agent.” | Done |
| G6 | Important | Equitable technology access. MIT highlights the $200/month divide between students who can afford premium AI subscriptions and those who cannot. The TFIS course assumes universal equal access to AI tools. | §3.3.7 | Add an equity-of-access dimension to the S6 maturity scan item on “Technology Provision.” Include a short discussion on institutional-provision models (Parley-style vs. BYO-subscription). | Done |
| G7 | Important | Continuous monitoring & metrics. MIT recommends tracking AI use, campus engagement, student satisfaction, and post-graduation outcomes. The TFIS course uses the EAI-CMM delta as its sole measurement instrument; no institutional feedback-loop layer exists. | §3.3.6 | Add a “Monitoring & Metrics” prompt (P11b) to S5–S6 transition. Include a recommendation in the guideline skeleton for annual review grounded in survey data. | Done |
| G8 | Important | AI-platform governance: logging, auditing, & privacy. MIT raises complex questions about institutional AI platforms (e.g., Parley): who can see student chat logs, how to handle mental-health flags, anonymization. The TFIS course covers PDPA as a data floor but not institutional AI-platform governance. | §3.3.8–3.3.9 | Add a “Platform Governance” section to the S6 policy-anatomy discussion. Include a brief case study on MIT’s Parley system and its privacy trade-offs. | Done |
| G9 | Supplementary | Environmental costs of AI. MIT discusses data-centre energy/water consumption and recommends auditing institutional AI footprints. The TFIS course does not address environmental sustainability of AI. | §3.3.10 | Add a footnote-level mention in S1’s evidence walk (2026 AI Index includes environmental data). Optional: include in the guideline’s principles component as a sustainability principle. | Not yet |
| G10 | Supplementary | Physical learning-space redesign. MIT recommends investing in in-person collaborative and AI-free physical spaces for labs and assessments. The TFIS course does not address the physical environment. | §3.1.7 | Add as a “bonus component” in the guideline skeleton: “Physical & Digital Learning Environments.” Optional discussion in S6. | Not yet |
| G11 | Supplementary | Curriculum governance process reform. MIT urges revising governance processes to allow rapid curricular exploration without multi-committee year-long reviews. The TFIS course does not address institutional curriculum-change mechanics. | §3.1.10, §3.3.1 | Optional: add a “Curriculum Governance” note to the guideline’s Ownership & Review component (Component 7). Low priority for this audience. | Not yet |
Of the 11 identified gaps, 8 are now implemented (all 3 Critical + 5 Important) in this revision. G9–G11 (Supplementary) remain as future work. The implementation touched 6 files: slide decks s1, s2, s5, s6; instruments (templates.md); prompts (prompt-handouts.md); and the capstone brief. Each gap’s recommended response has been translated into specific, actionable content changes — see the “Recommended Ad-Hoc Integration” cards below for the complete map of what was done and where.
For balance, the following course features address AI-in-education problems the MIT report under-treats or omits entirely:
| # | Course Strength | Why It Matters |
|---|---|---|
| R1 | AI-text-detector bias quantified (Liang et al., 61%+ false-positive) | MIT mentions detector caution but lacks the definitive evidence. The course’s S3 evidentiary core is stronger and more actionable. |
| R2 | Crutch-effect RCT (Bastani et al., −17% unassisted) | MIT discusses “cognitive surrender” at the conceptual level. The course offers a precise, design-relevant effect size that shapes practice. |
| R3 | 4D fluency framework (Delegation, Description, Discernment, Diligence) | MIT recommends AI literacy but provides no operational competency model. The 4Ds give instructors a teachable, assessable skill taxonomy. |
| R4 | EAI-CMM maturity instrument | MIT recommends tracking metrics (§3.3.6) but offers no instrument. The EAI-CMM measures movement across 5 pillars × 20 items. |
| R5 | Three-lane permission architecture (🔴🟡🟢) | MIT offers a policy menu with an example (Appendix B) but no shared design vocabulary. The three-lane pattern is teachable in 90 seconds and applicable across disciplines. |
| R6 | Malaysian regulatory grounding (PDPA, AI Governance Bill, MQA/MOHE slots) | MIT operates within US legal norms. The course’s S6–S7 Malaysian context is essential for this audience and entirely absent from the MIT report. |
| R7 | Detection-is-a-losing-regime (A4 axiom) | MIT advises caution with detectors but frames it as a tactical recommendation. The course makes the structural argument that the detection paradigm itself must be retired. |
| R8 | 15 model-agnostic prompts (P01–P15) with [BRACKET] values | MIT recommends training and support but provides no ready-to-use teaching tools. The course ships with an entire library of classroom-ready prompts. |
All items below have been implemented across 6 files.
S1 Campus Life row added to Threat/Gift Matrix (G1) · MIT report link in appendix (G9 ref)
S2 Card 9 added: “Faculty replaces UROP with AI agent” (G5) · Debrief note updated
S5 New slide 5b: “What if there were no grades?” discussion, MIT report §3.1.6 cited (G2)
S6 Components 4b, 8 added to policy anatomy · Component 6 + 7 amended (platform governance, monitoring metrics) · Mirror test added to gap analysis (G3, G4, G6, G7, G8)
templates.md D5 skeleton updated: Components 4b, 8 added; Components 6, 7 amended
prompt-handouts.md P05 strengthened with instructor-disclosure symmetry language · P11b (Monitoring & Metrics) added as new prompt
capstone-brief.md 9-component architecture documented · S6 row updated
templates.md Card 9 added to Ethics Triage scenario set
G9 — Environmental costs: Recommend footnote-level mention in S1 evidence walk. G10 — Physical learning-space redesign: Recommend bonus component in guideline skeleton. G11 — Curriculum governance reform: Low priority for this audience. These three supplementary gaps can be addressed in v1.2.
The 3 Critical and 5 Important gaps have been integrated into the course materials (this revision). The MIT report’s eight principles and three recommendation areas provide a strong external-validation touchpoint for the course’s opening and closing sessions. The MIT report has been added as a recommended reading resource across multiple appendix slides. Consider formally adding it to the source register HTML as a restricted source. Remaining supplementary gaps (G9–G11) can be addressed in v1.2.
09 · Schedule