Snapshot
Three-phase score, single page
PH PSG-GE 2026 Draft assessed across 14 international comparators on six benchmark lenses (Phase A), against the AI labor forecast for 2027-2032 graduates on five core courses (Phase B), and against the implementation infrastructure required to deliver the design (Phase C).
Phase A · Comparator
11/18
Mid-pack. Above ASEAN regional cohort, below Asian flagships.
Mid Position
Phase B · AI-era fit
9/15
Course-level composite. Course 5 most concerning.
Partial Fit
Phase C · Realism
8/18
Design strong on paper, infrastructure under-provisioned.
High Risk
PH Cohort Exposure
~33%
PH workers highly AI-exposed (IMF WP25/043, Feb 2025). BPO sector 89% high automation risk.
Acute
Phase A · Comparator Matrix
15 systems × 6 lenses, scored 0-3
Score shape matters more than total. PH escapes the bottom-tier civic-only cohort but holds the quantitative floor at floor. India CCFUP is the most relevant peer (shared SHS-strand handoff) and sits one step ahead.
| System | L1 | L2 | L3 | L4 | L5 | L6 | Total |
|---|---|---|---|---|---|---|---|
| HKU (Hong Kong) | 3 | 3 | 3 | 3 | 3 | 3 | 18 |
| NUS (Singapore) | 3 | 3 | 3 | 3 | 3 | 3 | 18 |
| NTU (Singapore) | 3 | 3 | 2 | 3 | 3 | 3 | 17 |
| MIT | 3 | 3 | 3 | 3 | 1 | 3 | 16 |
| Harvard | 2 | 3 | 2 | 2 | 3 | 3 | 15 |
| UTokyo Komaba | 2 | 3 | 3 | 2 | 1 | 3 | 14 |
| India CCFUP | 2 | 2 | 2 | 2 | 2 | 3 | 13 |
| UCL BASc (UK) | 2 | 3 | 2 | 2 | 1 | 3 | 13 |
| SNU (South Korea) | 1 | 2 | 3 | 2 | 1 | 2 | 11 |
| CUHK (Hong Kong) | 1 | 2 | 2 | 1 | 2 | 3 | 11 |
| PH PSG-GE 2026 Draft | 2 | 2 | 1 | 2 | 2 | 2 | 11 |
| Tsinghua (China) | 1 | 2 | 2 | 1 | 1 | 2 | 9 |
| Malaysia MPU | 0 | 1 | 0 | 1 | 1 | 1 | 4 |
| Indonesia MKWU | 0 | 1 | 0 | 0 | 1 | 1 | 3 |
| Vietnam MOET PT | 0 | 1 | 0 | 0 | 1 | 1 | 3 |
0 absent
1 weak / surface
2 moderate / single mandate
3 structurally embedded
L1 Future-of-work durability · L2 Cognitive density · L3 Quant floor · L4 AI-digital depth · L5 Civic-global balance · L6 SHS articulation
Phase A · Detail
PH against four most-relevant peers
PH plotted against India CCFUP (shared SHS-strand cohort, one step ahead), HKU (AI-mandate exemplar), NUS (two-pillar digital depth), and Indonesia MKWU (ASEAN regional baseline). The shape gap on L3 (Quantitative floor) and L4 (AI-digital depth) is the visible structural divergence.
Phase B · AI-Era Stress Test
Five core courses against the 2027-2032 labor forecast
Each course mapped against tasks AI is forecast to absorb, capabilities with rising premium, and the doing-to-overseeing skill-stack shift. Composite course-level fit: 9 of 15. Course 5 (Labor Education) is the most concerning gap relative to the cohort's labor-market exposure.
Course 1 · Professional Communication
3 units · CO3 portfolio with AI disclosure points right way; CO1 ILO1 and CO2 ILO2 teach what AI already does well
2 / 3
Course 2 · Global Trends & Emerging Technologies
3 units · Strong activity design; framing is descriptive (about AI) not operational (with/against AI). Reads 2018-vintage
2 / 3
Course 3 · Data, Evidence, and Ethics in a Knowledge Society
3 units · Overloaded. Stats 101 alone runs 6-8 units in HK/SG/US. Quant or themes optionality lets students exit without depth
2 / 3
Course 4 · Rizal and Philippine Studies
3 units · Hermeneutic capacity, civic adjudication, defended advocacy = AI-resistant + rising premium. Quietly future-proof
2 / 3
Course 5 · Labor Education
3 units · Pre-2023 industrial-relations vintage. Does not address BPO automation exposure, agentic-workplace labor rights, or career building under task substitution
1 / 3
Phase C · Drafted-Intent vs Deliverable-Reality
Six implementation lenses
Whether the system that has to deliver the draft can. Faculty pipeline, library, assessment infrastructure, and typology equity all score 1 of 3. The vulnerabilities concentrate in the same courses Phase A flagged (3 and 5), making the risk correlated rather than random.
L7
Faculty pipeline
Course 3 needs single-instructor profile fluent in stats + qual coding + research ethics + AI transparency. Course 2 needs 2024-2026 AI-tool fluency. No national bridging mechanism, training fund, or certification pathway in the draft.
1/3
High
L8
Library and learning resources
Real constraint is AI-tool access, statistical software, dataset licenses, digital-platform subscriptions. Top-tier autonomous HEIs absorb; SUCs and LUCs cannot. No infrastructure subsidy specified.
1/3
High
L9
Assessment and CQI infrastructure
GEO-CO-ILO three-level mapping, CQI artifacts, three-year review with tracer studies and employer feedback. Top-tier HEIs run assessment offices; mid- and lower-tier run paper-only OBE compliance. Standard set, infrastructure not provisioned.
1/3
High
L10
Typology equity
Autonomous HEIs may expand to 36 units; under-resourced left at 18-unit floor. Combined with unfunded faculty mandate, policy can plausibly widen the institutional quality gap rather than narrow it.
1/3
Med-High
L11
Transition timeline
AY 2026-27 pilot, AY 2027-28 full implementation, ~16 months. Comparable to India CCFUP (2.5 yr from policy) but materially thinner support scaffolding. Defensible but tight.
2/3
Medium
L12
QA and review logic
No course-by-course approval for institutional GE; three-year review with external sampling. High-trust regulatory model in low-trust environment. Catches drift eventually, does not prevent it during formative cohort.
2/3
Medium
Synthesis · The Shape of the Risk
Three shape problems that compound
The draft is not bad. It is thoughtful, structurally improved, and progressive relative to ASEAN regional peers. The risk lies in three coherent shape problems.
SHAPE PROBLEM 1
Quantitative floor held at floor
PH has elevated AI/civic/future-of-work above ASEAN regional peers but holds quant at ~3 units (Course 3, partial). MIT 72 units. Stanford 4 courses. NUS 8. HKU 9+. India 9. The single largest divergence from any peer above PH on the table. Structurally incongruent with an AI labor forecast where rising-premium skills are heavily quant/oversight-flavoured.
SHAPE PROBLEM 2
Disclosure governance, capability gap
Universal AI/integrity disclosure architecture is ethically sound and internally consistent. It produces the appearance of AI integration without the underlying skill formation. Graduates will know to disclose AI use; they will not necessarily know how to use AI well or evaluate AI outputs critically. Disclosure tells you what was used; it does not teach capability.
SHAPE PROBLEM 3
Designed, not resourced
The OBE-aligned, three-level-scaffolded, CQI-driven program at the autonomous HEI standard does not exist at the SUC/LUC and average-private-HEI standard, and is not provisioned. Most likely outcome at autonomous HEIs: partial success with localised innovation. Most likely outcome at the broader HEI population: rebadged compliance with the old curriculum under new labels.
Quiet Strengths
Three findings the lens scoring elevates
STRENGTH 1
Course 4 (Rizal) is among the most AI-era durable
Hermeneutic textual interpretation, defended civic advocacy, culturally-rooted judgment under public scrutiny. All rising-premium per WEF, Autor, McKinsey. The statutorily-mandated national-identity anchor turns out to be future-proof, not vestigial. Cheap upgrade available: read Rizal's media-power critique against contemporary AI-mediated civic discourse.
STRENGTH 2
Civic-global balance is genuinely well-calibrated
Few comparator systems achieve it. Tsinghua, Indonesia, Vietnam, Malaysia skew heavily national. MIT, UCL, UTokyo skew light on civic. PH joins the small group (with Harvard and HKU) of systems that hold both. The Rizal mandate provides national anchor without crowding out global content.
STRENGTH 3
OBE constructive-alignment architecture is sophisticated
GEO-CO-ILO mapping with three-level scaffolding (foundational/reinforcing/culminating), aligned TLAs, diagnostic-formative-culminating assessment progression, CQI evidence requirements. Best-practice OBE design. The architecture is sound. The constraint is the execution capability gap, not the design.
High-Leverage Interventions
What the gap analysis points to
INTERVENTION 1
Faculty-development annex with funded transition
Two-year national bridging program for current GE faculty into Course 2 and Course 3. Funded, certificated, pre-pilot completion required. AI-tool fluency, current quantitative pedagogy, and applied research ethics as core modules. Autonomous HEIs as training hubs. Without this, AY 2027-28 ships the courses but not the capability.
INTERVENTION 2
Course 3 unit reweighting or partner-course mechanism
Three units cannot carry data + descriptives + qualitative + ethics + AI transparency + executed inquiry + reporting. Three options: (a) expand to 6 units matching NUS Data Literacy alone; (b) split methodology and ethics into Course 3a/3b within institutional GE; (c) formally partner Course 3 with major-specific quant-methods courses for credit transfer. Option (c) likely most realistic.
INTERVENTION 3 · MOST TIMELY
Labor Education content modernisation directive
CHED clarifying memo updating Course 5 COs to include AI/automation labor-market reality without amending RA 11551. Sectoral exposure mapping (BPO, finance, legal, market research, content); reskilling pathways; labor rights in AI-augmented and agent-based workplaces; algorithmic management; career-building under task substitution. RA 11551 IRR signed late 2025/early 2026; window is live.