Learning Objectives
- Apply the c-ECO framework to a real or realistic AI Systems engagement.
- Translate sector data into TFP variables and prudential classifications.
- Design pre-threshold governance mechanisms for acceptable error, drift, and harm boundaries, human oversight capacity limits, data integrity and bias thresholds.
- Present systemic analysis to non-technical stakeholders.
- Produce a professional-grade CSAM packet suitable for controlled institutional coordination.
Field Project: The c-ECO AI Systems Stability Assessment
Your Role: c-ECO Fellow assigned to conduct a supervised Systemic Stability Assessment for an automated decision system, model deployment, AI infrastructure dependency, algorithmic governance process, or compute-intensive platform with systemic risk or public-interest exposure.
Project Scope: Produce a professional-grade CSAM packet that defines system boundary, signals, thresholds, actors, reversibility exposure, and institutional translation options. This is a Fellowship analytical exercise and does not authorize independent deployment.
- System boundary identification: models, training data, automated decisions, computational infrastructure, human oversight, governance controls, public impact, and institutional accountability.
- Safe Operating Space boundary identification for the selected subsystem.
- Current Position assessment with uncertainty treatment.
- Trajectory analysis using historical, reported, modeled, or field-linked data.
- Reversibility Liquidity evaluation of technical, financial, institutional, and temporal capacity.
- Band classification: Green, Amber, Red/Safe Mode, or Black/Restoration First.
- Contractual and governance recommendations for pre-threshold intervention.
- Implementation roadmap for supervised institutional use where authorized.
Select Your AI Systems Subsystem
model drift, performance degradation, and unexplained decision variance.
data bias, representational failure, and dataset obsolescence.
compute and energy burden acceleration.
automation dependency and human oversight erosion.
model developers and deployers, data providers and infrastructure operators, affected users and communities.
model drift Safe Mode clauses, human oversight covenants, data integrity schedules.
Required Deliverables
Format: 12–18 page professional report.
- Executive Summary for cohort or institutional coordination.
- System boundary, actor map, and exposure logic.
- Data audit: sources, coverage, gaps, uncertainty, and monitoring frequency.
- TFP variables: P, ΔV, σ, Lr with transparent assumptions.
- Prudential classification and escalation conditions.
- Safe Mode or Restoration First implications where applicable.
Format: Case-Specific Analytical Mandate with scope, evidence, limits, and intervention logic.
- Case purpose and institutional boundary.
- Threshold map and signal architecture.
- Actor duties and information dependencies.
- Confidentiality, data, and methodological limitations.
Format: 15-minute presentation plus questions.
Translate technical findings into clear governance consequences for model developers and deployers, data providers and infrastructure operators, affected users and communities, risk, legal, compliance, and oversight teams.
Format: 3-page reflection on where ordinary compliance would arrive too late, how c-ECO changes responsibility, and how the Fellow preserves institutional boundaries.
Assessment Rubric
| Criterion | Excellent | Good | Satisfactory | Needs Work |
|---|---|---|---|---|
| Technical Accuracy | Precise TFP logic, strong evidence treatment, and credible sector assumptions. | Mostly correct with minor gaps. | Basic variables, limited sensitivity analysis. | Major conceptual or measurement errors. |
| Systemic Interpretation | Clearly distinguishes incident, trajectory, and reversibility loss. | Good analysis with some generic passages. | Identifies risk but weakly links it to thresholds. | Treats case as ordinary compliance. |
| CSAM Quality | Scope, actors, evidence, limits, and intervention logic are coherent and usable. | Strong structure with minor omissions. | Basic mandate, incomplete operational logic. | Mandate unclear or not case-specific. |
| Professional Communication | Board-ready, concise, and disciplined. | Professional with minor polish needed. | Understandable but not executive-ready. | Unclear or overly generic. |
Project Timeline
Week 1:
- Days 1–2: Select subsystem, define case boundary, gather data.
- Days 3–4: Calculate P, ΔV, σ, and Lr; identify preliminary band classification.
- Days 5–7: Draft Technical Assessment Report and identify missing monitoring data.
Week 2:
- Days 8–10: Complete Safe Mode / Restoration First analysis and CSAM draft.
- Days 11–12: Prepare institutional briefing.
- Day 13: Submit all deliverables through cohort coordination.
- Day 14: Live presentation to instructor and peers.
Resources and Support
Technical Resources
- Modules 1–4 notes and TFP variable templates.
- Sector data sources: NIST AI Risk Management Framework; OECD AI principles; ISO AI governance materials.
- Office hours: one 30-minute consultation during Week 1.
Governance and Instrument Sources
- Model drift Safe Mode clauses.
- Human oversight covenants.
- Data integrity schedules.
- Algorithmic impact escalation triggers.
- Compute burden and dependency protocols.