Skill Change Index (SCI)
Skill Change Index (SCI)
McKinsey's measure (GCC Philippines Summit 2026 (PHx)) of how much AI reprices the skills demanded by a role — the degree to which a given skill's relevance rises or falls as AI automates parts of the work. Computed across ~6,800 skills (sources: Lightcast, US BLS, McKinsey).
The shape (illustrative: Account Executives, median SCI 28%)
- Decreasing in relevance (high automation exposure): spreadsheet software, CRM software, lead generation, sales-territory management — routine, tool-bound, codifiable work.
- Increasing in relevance: influencing, negotiation, forecasting, customer-relationship management, judgment — human-, hybrid-, and context-heavy work.
The headline
"AI will not just reduce work — it will reprice skills." The Philippines must shift talent from routine execution to judgment, influence, and AI-enabled problem-solving. The winners build skill corridors — defined migration paths that move people from compressing roles into rising ones.
2026-06-13 — Agent skills are mostly re-application, not greenfield
The 7 Skills You Need to Build AI Agents (IBM Technology) (Bri Kopecki) reinforces the SCI thesis from the AI-engineering side. Of her seven production-agent skills, five (system design, reliability, security, evaluation/observability, and arguably retrieval) are pre-existing backend / distributed-systems / security disciplines re-pointed at LLM-driven systems — not net-new competencies. In SCI terms most "agent" skills score as low skill-change / re-application of existing expertise, not greenfield reskilling: the platform/SRE/security engineer already speaks the language. That sharpens the repricing story — the relevance of those backend skills rises as they get re-aimed at agents, rather than being displaced.
Connections
The talent-side mechanism behind the Headcount-to-Value Pivot and the Frontier GCC talent pyramid. Directly feeds the user's Designing IT Roles for an AI Era (Talent Strategy POV) and DRAG for AI Upskilling at Manila IT Site. Compare the de-skilling risk in Designing AI Products That Don't De-Skill Users and Cognitive Offloading.