From 5cf78a31ae96480f4c2fbd2915d6c669006eb664 Mon Sep 17 00:00:00 2001 From: Fabrice Nemo Date: Wed, 16 Sep 2026 12:05:54 +0200 Subject: [PATCH 1/6] Reorder Weeks 7 to 9 schedule --- content/pages/changelog.md | 8 +- .../student/project-overview.md | 2 +- .../project-documentation/student/rubrics.md | 2 +- ..._the_cure_become_worse_than_the_disease.md | 44 +++++------ ...ek_08_project_studio_and_midterm_review.md | 49 ++++++------ ..._09_what_does_it_mean_to_trust_a_number.md | 74 +++++++++---------- 6 files changed, 91 insertions(+), 88 deletions(-) diff --git a/content/pages/changelog.md b/content/pages/changelog.md index 2a3f3aa..adf8f63 100644 --- a/content/pages/changelog.md +++ b/content/pages/changelog.md @@ -5,10 +5,14 @@ nav_title: "Changelog" page_heading: "Changelog" sidebar_group: "Course" order: 1 -last_edited: "2026-09-15" -change_note: "Changed weekly content release to 13:00 Europe/Zurich." +last_edited: "2026-09-16" +change_note: "Reordered Weeks 7-9 and moved midterm presentations to Week 9." --- +## 16/09/2026 + +- Reordered Weeks 7-9: The Fragility of Accuracy is now Week 7, Advice at the Wrong Time is Week 8, and midterm presentations are now Week 9. Midterm submissions remain due in Week 8. + ## 15/09/2026 - Updated the weekly content release to 13:00 Europe/Zurich. diff --git a/content/project-documentation/student/project-overview.md b/content/project-documentation/student/project-overview.md index 91f3b52..cb3e13c 100644 --- a/content/project-documentation/student/project-overview.md +++ b/content/project-documentation/student/project-overview.md @@ -47,7 +47,7 @@ Deadlines are set at 21:59 on Sundays. | September 16, 21:59 | Wednesday of Week 2 | **Deadline for project applications** (apply on Moodle) | | September 17 | Thursday of Week 2 | **Announcement of project assignments** | | September 27, 21:59 | Sunday of Week 3 | Milestone 1: **Technical design and reproducible repository skeleton** | -| November 4, 13:15 | Wednesday of Week 8 | **Proof of concept: Oral presentations** | +| November 11, 13:15 | Wednesday of Week 9 | **Proof of concept: Oral presentations** | | November 8, 21:59 | Sunday of Week 8 | Milestone 2: **Proof of concept: Submissions** | | November 23-27 | Week 11 | **30-minute team checkpoint with a TA** | | December 16, 13:15 | Wednesday of Week 14 | **Final presentations** | diff --git a/content/project-documentation/student/rubrics.md b/content/project-documentation/student/rubrics.md index c9907bf..ffe45cd 100644 --- a/content/project-documentation/student/rubrics.md +++ b/content/project-documentation/student/rubrics.md @@ -17,7 +17,7 @@ Course-grade weights have priority. The project-related components account for 8 |---|---:|---:|---| | Milestone 1: Technical Design and Reproducible Repository | 10% | 12.5% | Technical design and reproducible repository skeleton | | Milestone 2: Proof Of Concept | 20% | 25% | Proof of concept technical checkpoint | -| Midterm oral presentation with live demo | 10% | 12.5% | Week 8 oral presentation and live demo | +| Midterm oral presentation with live demo | 10% | 12.5% | Week 9 oral presentation and live demo | | Final oral presentation with live demo | 10% | 12.5% | Final submission | | Project report and code | 30% | 37.5% | Final submission | diff --git a/content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md b/content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md index 9d6a4b2..341e6f9 100644 --- a/content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md +++ b/content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md @@ -1,24 +1,24 @@ --- page_id: "Week_07" -page_title: "Week 7: Advice at the Wrong Time" -nav_title: "Week 7 - Advice at the Wrong Time" +page_title: "Week 7: The Fragility of Accuracy" +nav_title: "Week 7 - Fragility of Accuracy" sidebar_group: "Block II - High-Stakes Decisions" order: 7 week: 7 lecture_date: "2026-10-28" -theme: "Advice at the Wrong Time" -context_lecture: "Why Doctors Ignore Good Advice" +theme: "The Fragility of Accuracy" +context_lecture: "How We Learned to Count What We Were Losing" slides_pdf: "" -engineering_lecture: "Integration Without Interruption" +engineering_lecture: "When 99% Accuracy Is Dangerous" --- ## Overview **Block II: High-Stakes Decisions** (Weeks 5-8) -**Central question:** When does the cure become worse than the disease? +**Central question:** What does it mean to trust a number? -**Big idea:** Implementation inside clinical workflows without creating new harm. +**Big idea:** Why published performance can be misleading and what valid evaluation requires. **Block focus:** Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. @@ -26,19 +26,19 @@ engineering_lecture: "Integration Without Interruption" After this lecture, you should be able to: -- Explain why a correct recommendation can still make care worse. -- Identify workflow, timing, alert fatigue, and accountability risks in decision support. -- Compare interruptive and non-interruptive integration patterns. -- Translate workflow analysis into safer AI system behavior. +- Explain why a single performance number can hide clinically important failure modes. +- Identify leakage, poor validation design, and subgroup failure in AI evaluation. +- Compare internal, external, temporal, and population-level validation questions. +- Translate trustworthy evaluation into concrete reporting and engineering practices. ## Core Ideas and Case Anchor **Key concepts** -- Evidence-based medicine -- Clinical workflows -- Workflow integration -- Decision support safety +- Validation +- Benchmarking +- Clinical AI evaluation +- Misleading performance metrics **Context themes in this block** @@ -57,21 +57,21 @@ After this lecture, you should be able to: **Case anchor** -Placeholder: add the workflow or alert-fatigue case discussed in the lecture. +Placeholder: add the evaluation failure, benchmark, or clinical AI validation case discussed in the lecture. ## Engineering Takeaways -- Placeholder: summarize where the system should fit into the user's workflow. -- Placeholder: identify one condition under which the system should remain silent. -- Placeholder: state how alerts, logging, or handoff should be designed to avoid new harm. +- Placeholder: summarize which metrics students should report beyond a headline accuracy score. +- Placeholder: identify one validation split or subgroup analysis that should be added to projects. +- Placeholder: state how results should be communicated so limitations are visible. ## Project Reflection and Further Reading **Reflection questions** -- At what moment would your system ask for attention, and why then? -- What would happen if every recommendation from your system were technically correct but badly timed? -- What user action should your system make easier, not harder? +- Which performance number in your project could be misleading? +- What subgroup, site, or time period might expose a hidden failure? +- What result would make your team stop or redesign the model? **Further reading** diff --git a/content/weeks/week_08_project_studio_and_midterm_review.md b/content/weeks/week_08_project_studio_and_midterm_review.md index 660c5de..65940b3 100644 --- a/content/weeks/week_08_project_studio_and_midterm_review.md +++ b/content/weeks/week_08_project_studio_and_midterm_review.md @@ -1,43 +1,44 @@ --- page_id: "Week_08" -page_title: "Week 8: Midterm Submission and Presentations" -nav_title: "Week 8 - Midterm" +page_title: "Week 8: Advice at the Wrong Time" +nav_title: "Week 8 - Advice at the Wrong Time" sidebar_group: "Block II - High-Stakes Decisions" order: 8 week: 8 lecture_date: "2026-11-04" -theme: "Midterm Submission and Presentations" -context_lecture: "Midterm presentations" +theme: "Advice at the Wrong Time" +context_lecture: "Why Doctors Ignore Good Advice" slides_pdf: "" -engineering_lecture: "Midterm presentations" +engineering_lecture: "Integration Without Interruption" --- ## Overview **Block II: High-Stakes Decisions** (Weeks 5-8) -**Central question:** Is the system design technically and contextually defensible? +**Central question:** When does the cure become worse than the disease? -**Big idea:** Consolidate the problem, design, evidence plan, and implementation path. +**Big idea:** Implementation inside clinical workflows without creating new harm. **Block focus:** Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. ## Learning Objectives -After this session, you should be able to: +After this lecture, you should be able to: -- Present the current project design clearly and defensibly. -- Explain the main technical, contextual, and evaluation risks in the project. -- Identify what evidence is still missing before the proof of concept. -- Translate feedback into concrete next engineering steps. +- Explain why a correct recommendation can still make care worse. +- Identify workflow, timing, alert fatigue, and accountability risks in decision support. +- Compare interruptive and non-interruptive integration patterns. +- Translate workflow analysis into safer AI system behavior. ## Core Ideas and Case Anchor **Key concepts** -- Midterm presentation -- Technical design review -- Project checkpoint +- Evidence-based medicine +- Clinical workflows +- Workflow integration +- Decision support safety **Context themes in this block** @@ -56,23 +57,23 @@ After this session, you should be able to: **Case anchor** -Placeholder: summarize the main project-review pattern or example discussed during the session. +Placeholder: add the workflow or alert-fatigue case discussed in the lecture. ## Engineering Takeaways -- Placeholder: summarize what distinguishes a defensible technical design from a promising idea. -- Placeholder: identify one implementation risk that teams should reduce before the next milestone. -- Placeholder: state how teams should document design changes after receiving feedback. +- Placeholder: summarize where the system should fit into the user's workflow. +- Placeholder: identify one condition under which the system should remain silent. +- Placeholder: state how alerts, logging, or handoff should be designed to avoid new harm. ## Project Reflection and Further Reading **Reflection questions** -- What is the most important design change your team should make after the review? -- Which risk should be tested first in the proof of concept? -- What evidence would make your project scope simpler or more credible? +- At what moment would your system ask for attention, and why then? +- What would happen if every recommendation from your system were technically correct but badly timed? +- What user action should your system make easier, not harder? **Further reading** -- Core: To be added by the lecturer after the session. -- Optional: To be added by the lecturer after the session. +- Core: To be added by the lecturer after the lecture. +- Optional: To be added by the lecturer after the lecture. diff --git a/content/weeks/week_09_what_does_it_mean_to_trust_a_number.md b/content/weeks/week_09_what_does_it_mean_to_trust_a_number.md index 00309b2..7180258 100644 --- a/content/weeks/week_09_what_does_it_mean_to_trust_a_number.md +++ b/content/weeks/week_09_what_does_it_mean_to_trust_a_number.md @@ -1,80 +1,78 @@ --- page_id: "Week_09" -page_title: "Week 9: The Fragility of Accuracy" -nav_title: "Week 9 - Fragility of Accuracy" -sidebar_group: "Block III - Trustworthy Evidence" +page_title: "Week 9: Midterm Presentations" +nav_title: "Week 9 - Midterm Presentations" +sidebar_group: "Block II - High-Stakes Decisions" order: 9 week: 9 lecture_date: "2026-11-11" -theme: "The Fragility of Accuracy" -context_lecture: "How We Learned to Count What We Were Losing" +theme: "Midterm Presentations" +context_lecture: "Midterm presentations" slides_pdf: "" -engineering_lecture: "When 99% Accuracy Is Dangerous" +engineering_lecture: "Midterm presentations" --- ## Overview -**Block III: Trustworthy Evidence** (Weeks 9-13) +**Block II: High-Stakes Decisions** (Weeks 5-9) -**Central question:** What does it mean to trust a number? +**Central question:** Is the system design technically and contextually defensible? -**Big idea:** Why published performance can be misleading and what valid evaluation requires. +**Big idea:** Consolidate the problem, design, evidence plan, and implementation path. -**Block focus:** Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. +**Block focus:** Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. ## Learning Objectives -After this lecture, you should be able to: +After this session, you should be able to: -- Explain why a single performance number can hide clinically important failure modes. -- Identify leakage, poor validation design, and subgroup failure in AI evaluation. -- Compare internal, external, temporal, and population-level validation questions. -- Translate trustworthy evaluation into concrete reporting and engineering practices. +- Present the current project design clearly and defensibly. +- Explain the main technical, contextual, and evaluation risks in the project. +- Identify what evidence is still missing before the proof of concept. +- Translate feedback into concrete next engineering steps. ## Core Ideas and Case Anchor **Key concepts** -- Validation -- Benchmarking -- Clinical AI evaluation -- Misleading performance metrics +- Midterm presentation +- Technical design review +- Project checkpoint **Context themes in this block** -- Clinical trials -- Implementation science -- Regulation and governance -- Health systems +- Clinical reasoning +- Diagnosis and triage +- Evidence-based medicine +- Clinical workflows **Engineering themes in this block** -- Validation -- Benchmarking -- Safety -- Deployment -- Monitoring -- Technical communication +- Data analysis +- Bayesian reasoning +- Causal inference +- RAG +- Clinical decision support **Case anchor** -Placeholder: add the evaluation failure, benchmark, or clinical AI validation case discussed in the lecture. +Placeholder: summarize the main project-review pattern or example discussed during the session. ## Engineering Takeaways -- Placeholder: summarize which metrics students should report beyond a headline accuracy score. -- Placeholder: identify one validation split or subgroup analysis that should be added to projects. -- Placeholder: state how results should be communicated so limitations are visible. +- Placeholder: summarize what distinguishes a defensible technical design from a promising idea. +- Placeholder: identify one implementation risk that teams should reduce before the next milestone. +- Placeholder: state how teams should document design changes after receiving feedback. ## Project Reflection and Further Reading **Reflection questions** -- Which performance number in your project could be misleading? -- What subgroup, site, or time period might expose a hidden failure? -- What result would make your team stop or redesign the model? +- What is the most important design change your team should make after the review? +- Which risk should be tested first in the proof of concept? +- What evidence would make your project scope simpler or more credible? **Further reading** -- Core: To be added by the lecturer after the lecture. -- Optional: To be added by the lecturer after the lecture. +- Core: To be added by the lecturer after the session. +- Optional: To be added by the lecturer after the session. From 201afc099554d12f66d9bdb6c61b217c1ad1b3b1 Mon Sep 17 00:00:00 2001 From: Fabrice Nemo Date: Wed, 16 Sep 2026 12:11:42 +0200 Subject: [PATCH 2/6] Rename reordered week files and note presentation date change --- content/project-documentation/student/project-overview.md | 4 ++-- ...an_the_disease.md => week_07_the_fragility_of_accuracy.md} | 0 ..._midterm_review.md => week_08_advice_at_the_wrong_time.md} | 0 ..._to_trust_a_number.md => week_09_midterm_presentations.md} | 0 4 files changed, 2 insertions(+), 2 deletions(-) rename content/weeks/{week_07_when_does_the_cure_become_worse_than_the_disease.md => week_07_the_fragility_of_accuracy.md} (100%) rename content/weeks/{week_08_project_studio_and_midterm_review.md => week_08_advice_at_the_wrong_time.md} (100%) rename content/weeks/{week_09_what_does_it_mean_to_trust_a_number.md => week_09_midterm_presentations.md} (100%) diff --git a/content/project-documentation/student/project-overview.md b/content/project-documentation/student/project-overview.md index cb3e13c..36139e0 100644 --- a/content/project-documentation/student/project-overview.md +++ b/content/project-documentation/student/project-overview.md @@ -1,6 +1,6 @@ --- -last_edited: "2026-09-10" -change_note: "Project deadline policy clarified: deadlines are Sundays at 21:59. Added mandatory weekly decision-log and TA-update expectations." +last_edited: "2026-09-16" +change_note: "Proof-of-concept oral presentations moved to Week 9 (November 11)." --- # Course Project Overview diff --git a/content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md b/content/weeks/week_07_the_fragility_of_accuracy.md similarity index 100% rename from content/weeks/week_07_when_does_the_cure_become_worse_than_the_disease.md rename to content/weeks/week_07_the_fragility_of_accuracy.md diff --git a/content/weeks/week_08_project_studio_and_midterm_review.md b/content/weeks/week_08_advice_at_the_wrong_time.md similarity index 100% rename from content/weeks/week_08_project_studio_and_midterm_review.md rename to content/weeks/week_08_advice_at_the_wrong_time.md diff --git a/content/weeks/week_09_what_does_it_mean_to_trust_a_number.md b/content/weeks/week_09_midterm_presentations.md similarity index 100% rename from content/weeks/week_09_what_does_it_mean_to_trust_a_number.md rename to content/weeks/week_09_midterm_presentations.md From 45ef03c7c92d9db1f4f8ee9ffef805f78ddc3a03 Mon Sep 17 00:00:00 2001 From: Fabrice Nemo Date: Wed, 16 Sep 2026 12:14:09 +0200 Subject: [PATCH 3/6] Keep previous project overview edit notice --- content/project-documentation/student/project-overview.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/content/project-documentation/student/project-overview.md b/content/project-documentation/student/project-overview.md index 36139e0..bd7ef50 100644 --- a/content/project-documentation/student/project-overview.md +++ b/content/project-documentation/student/project-overview.md @@ -5,6 +5,8 @@ change_note: "Proof-of-concept oral presentations moved to Week 9 (November 11). # Course Project Overview +

Edited on 10/09: Added mandatory weekly decision-log and TA-update expectations.

+ The course project is a **semester-long team project** in which you will **design, build, evaluate, and document an AI system** for **global health or humanitarian response**. The goal is not only to train a model or build an interface. We want to make you **build a credible prototype of an AI system** that could support **decision making, communication, translation, risk analysis, or clinical work** in a **constrained and high-stakes setting**. From 5425b1138016665a9c90d6eb41e75c3399416252 Mon Sep 17 00:00:00 2001 From: Fabrice Nemo Date: Wed, 16 Sep 2026 12:20:57 +0200 Subject: [PATCH 4/6] Align block ranges with reordered schedule --- content/pages/home.md | 8 +++---- .../week_07_the_fragility_of_accuracy.md | 23 ++++++++++--------- .../weeks/week_09_midterm_presentations.md | 2 +- ..._whether_an_intervention_actually_works.md | 2 +- ...11_what_happens_when_an_ai_system_fails.md | 2 +- ..._who_is_responsible_when_ai_causes_harm.md | 2 +- ...ntions_rarely_reach_those_who_need_them.md | 2 +- 7 files changed, 21 insertions(+), 20 deletions(-) diff --git a/content/pages/home.md b/content/pages/home.md index df87368..f7effc7 100644 --- a/content/pages/home.md +++ b/content/pages/home.md @@ -28,8 +28,8 @@ The course is organized into **three thematic blocks**. Each teaching week pairs | Block | Weeks | Goal | |---|---:|---| | **Volatile Contexts** | 1-4 | Understand the environments in which AI must operate, and why technology often fails in humanitarian and clinical settings. | -| **High-stakes Decisions** | 5-7 | Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. | -| **Trustworthy Evidence** | 9-13 | Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. | +| **High-stakes Decisions** | 5-8 | Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. | +| **Trustworthy Evidence** | 7, 10-13 | Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. | {{ weeks_table }} @@ -55,7 +55,7 @@ Goal: Understand the environments in which AI must operate, and why technology o - Reproducibility - Edge computing -### **Block II: High-stakes Decisions** (Weeks 5-7) +### **Block II: High-stakes Decisions** (Weeks 5-8) Goal: Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. @@ -74,7 +74,7 @@ Goal: Understand how clinicians make decisions under uncertainty and how AI can - RAG - Clinical decision support -### **Block III: Trustworthy Evidence** (Weeks 9-13) +### **Block III: Trustworthy Evidence** (Weeks 7, 10-13) Goal: Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. diff --git a/content/weeks/week_07_the_fragility_of_accuracy.md b/content/weeks/week_07_the_fragility_of_accuracy.md index 341e6f9..adbfae7 100644 --- a/content/weeks/week_07_the_fragility_of_accuracy.md +++ b/content/weeks/week_07_the_fragility_of_accuracy.md @@ -14,13 +14,13 @@ engineering_lecture: "When 99% Accuracy Is Dangerous" ## Overview -**Block II: High-Stakes Decisions** (Weeks 5-8) +**Block III: Trustworthy Evidence** (Weeks 7, 10-13) **Central question:** What does it mean to trust a number? **Big idea:** Why published performance can be misleading and what valid evaluation requires. -**Block focus:** Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. +**Block focus:** Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. ## Learning Objectives @@ -42,18 +42,19 @@ After this lecture, you should be able to: **Context themes in this block** -- Clinical reasoning -- Diagnosis and triage -- Evidence-based medicine -- Clinical workflows +- Clinical trials +- Implementation science +- Regulation and governance +- Health systems **Engineering themes in this block** -- Data analysis -- Bayesian reasoning -- Causal inference -- RAG -- Clinical decision support +- Validation +- Benchmarking +- Safety +- Deployment +- Monitoring +- Technical communication **Case anchor** diff --git a/content/weeks/week_09_midterm_presentations.md b/content/weeks/week_09_midterm_presentations.md index 7180258..1c253ef 100644 --- a/content/weeks/week_09_midterm_presentations.md +++ b/content/weeks/week_09_midterm_presentations.md @@ -14,7 +14,7 @@ engineering_lecture: "Midterm presentations" ## Overview -**Block II: High-Stakes Decisions** (Weeks 5-9) +**Project checkpoint** (Week 9) **Central question:** Is the system design technically and contextually defensible? diff --git a/content/weeks/week_10_how_do_we_know_whether_an_intervention_actually_works.md b/content/weeks/week_10_how_do_we_know_whether_an_intervention_actually_works.md index eb0c9e5..144604e 100644 --- a/content/weeks/week_10_how_do_we_know_whether_an_intervention_actually_works.md +++ b/content/weeks/week_10_how_do_we_know_whether_an_intervention_actually_works.md @@ -14,7 +14,7 @@ engineering_lecture: "Causal Inference: Beyond Correlation" ## Overview -**Block III: Trustworthy Evidence** (Weeks 9-13) +**Block III: Trustworthy Evidence** (Weeks 7, 10-13) **Central question:** How do we know whether an intervention actually works? diff --git a/content/weeks/week_11_what_happens_when_an_ai_system_fails.md b/content/weeks/week_11_what_happens_when_an_ai_system_fails.md index da89e49..dca9d2e 100644 --- a/content/weeks/week_11_what_happens_when_an_ai_system_fails.md +++ b/content/weeks/week_11_what_happens_when_an_ai_system_fails.md @@ -14,7 +14,7 @@ engineering_lecture: "Uncertainty as Architecture" ## Overview -**Block III: Trustworthy Evidence** (Weeks 9-13) +**Block III: Trustworthy Evidence** (Weeks 7, 10-13) **Central question:** What happens when an AI system fails? diff --git a/content/weeks/week_12_who_is_responsible_when_ai_causes_harm.md b/content/weeks/week_12_who_is_responsible_when_ai_causes_harm.md index 6b62bf8..a5f8611 100644 --- a/content/weeks/week_12_who_is_responsible_when_ai_causes_harm.md +++ b/content/weeks/week_12_who_is_responsible_when_ai_causes_harm.md @@ -14,7 +14,7 @@ engineering_lecture: "Governing What You Cannot Fully Understand" ## Overview -**Block III: Trustworthy Evidence** (Weeks 9-13) +**Block III: Trustworthy Evidence** (Weeks 7, 10-13) **Central question:** Who is responsible when AI causes harm? diff --git a/content/weeks/week_13_why_do_effective_interventions_rarely_reach_those_who_need_them.md b/content/weeks/week_13_why_do_effective_interventions_rarely_reach_those_who_need_them.md index 53d2b0e..5d3b5f0 100644 --- a/content/weeks/week_13_why_do_effective_interventions_rarely_reach_those_who_need_them.md +++ b/content/weeks/week_13_why_do_effective_interventions_rarely_reach_those_who_need_them.md @@ -14,7 +14,7 @@ engineering_lecture: "Shipping to Scale: The MLOps of Global Health AI" ## Overview -**Block III: Trustworthy Evidence** (Weeks 9-13) +**Block III: Trustworthy Evidence** (Weeks 7, 10-13) **Central question:** Why do effective interventions rarely reach those who need them? From 5343e47a49375a1b14f33622ba1779c7c5707bd8 Mon Sep 17 00:00:00 2001 From: Fabrice Nemo Date: Wed, 16 Sep 2026 12:25:35 +0200 Subject: [PATCH 5/6] Clarify non-contiguous high-stakes block weeks --- content/pages/home.md | 4 ++-- ..._05_why_has_the_best_medical_ai_never_treated_a_patient.md | 2 +- .../week_06_how_do_experts_reliably_deceive_themselves.md | 2 +- content/weeks/week_08_advice_at_the_wrong_time.md | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/content/pages/home.md b/content/pages/home.md index f7effc7..080d6f7 100644 --- a/content/pages/home.md +++ b/content/pages/home.md @@ -28,7 +28,7 @@ The course is organized into **three thematic blocks**. Each teaching week pairs | Block | Weeks | Goal | |---|---:|---| | **Volatile Contexts** | 1-4 | Understand the environments in which AI must operate, and why technology often fails in humanitarian and clinical settings. | -| **High-stakes Decisions** | 5-8 | Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. | +| **High-stakes Decisions** | 5, 6, 8 | Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. | | **Trustworthy Evidence** | 7, 10-13 | Design, evaluate, and deploy AI systems that are safe, effective, and ready for real-world use. | {{ weeks_table }} @@ -55,7 +55,7 @@ Goal: Understand the environments in which AI must operate, and why technology o - Reproducibility - Edge computing -### **Block II: High-stakes Decisions** (Weeks 5-8) +### **Block II: High-stakes Decisions** (Weeks 5, 6, 8) Goal: Understand how clinicians make decisions under uncertainty and how AI can safely support human expertise. diff --git a/content/weeks/week_05_why_has_the_best_medical_ai_never_treated_a_patient.md b/content/weeks/week_05_why_has_the_best_medical_ai_never_treated_a_patient.md index 3bd61fd..d503b05 100644 --- a/content/weeks/week_05_why_has_the_best_medical_ai_never_treated_a_patient.md +++ b/content/weeks/week_05_why_has_the_best_medical_ai_never_treated_a_patient.md @@ -14,7 +14,7 @@ engineering_lecture: "From Rules to Reasoning" ## Overview -**Block II: High-Stakes Decisions** (Weeks 5-8) +**Block II: High-Stakes Decisions** (Weeks 5, 6, 8) **Central question:** Why has the best medical AI never treated a patient? diff --git a/content/weeks/week_06_how_do_experts_reliably_deceive_themselves.md b/content/weeks/week_06_how_do_experts_reliably_deceive_themselves.md index 72ee0e1..086b5ec 100644 --- a/content/weeks/week_06_how_do_experts_reliably_deceive_themselves.md +++ b/content/weeks/week_06_how_do_experts_reliably_deceive_themselves.md @@ -14,7 +14,7 @@ engineering_lecture: "Debiasing the Machine" ## Overview -**Block II: High-Stakes Decisions** (Weeks 5-8) +**Block II: High-Stakes Decisions** (Weeks 5, 6, 8) **Central question:** How do experts reliably deceive themselves, and can machines do better? diff --git a/content/weeks/week_08_advice_at_the_wrong_time.md b/content/weeks/week_08_advice_at_the_wrong_time.md index 65940b3..70b7ac3 100644 --- a/content/weeks/week_08_advice_at_the_wrong_time.md +++ b/content/weeks/week_08_advice_at_the_wrong_time.md @@ -14,7 +14,7 @@ engineering_lecture: "Integration Without Interruption" ## Overview -**Block II: High-Stakes Decisions** (Weeks 5-8) +**Block II: High-Stakes Decisions** (Weeks 5, 6, 8) **Central question:** When does the cure become worse than the disease? From 4dddd8b2d987e16e9af534692517456be86e755f Mon Sep 17 00:00:00 2001 From: fabnemEPFL <117652591+fabnemEPFL@users.noreply.github.com> Date: Wed, 16 Sep 2026 12:34:16 +0200 Subject: [PATCH 6/6] Change sidebar group from Block II to Block III for week 7 --- content/weeks/week_07_the_fragility_of_accuracy.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/weeks/week_07_the_fragility_of_accuracy.md b/content/weeks/week_07_the_fragility_of_accuracy.md index adbfae7..6de3b21 100644 --- a/content/weeks/week_07_the_fragility_of_accuracy.md +++ b/content/weeks/week_07_the_fragility_of_accuracy.md @@ -2,7 +2,7 @@ page_id: "Week_07" page_title: "Week 7: The Fragility of Accuracy" nav_title: "Week 7 - Fragility of Accuracy" -sidebar_group: "Block II - High-Stakes Decisions" +sidebar_group: "Block III - Trustworthy Evidence" order: 7 week: 7 lecture_date: "2026-10-28"