ENGLISH RECORD 80
How AI Is Changing Work
Credible forecasts state assumptions, task definitions and time horizons.01 · WHAT IT IS
Overview
AI may automate tasks, augment workers and reorganize occupations without mapping neatly to whole-job replacement.
02 · ORIGINAL TEXT AND EARLIEST SOURCE
What can be traced?
03 · HISTORICAL BACKGROUND
The world behind the text
64 · AI & LABOUR OBSERVATORY
How AI Is Changing Work
Tasks change before occupations disappear
Instead of forecasting a list of vanishing jobs, this dossier tracks which tasks are exposed, where AI is actually used, and how employment, wages and entry routes change. It keeps observations from 2022–2026 separate from projections to 2030.Exposure is not job loss00 · DEFINITIONS
Similar-looking numbers answer different questions
Exposure, use, automation and employment outcomes must not be collapsed into one risk score.Exposure
Technical potential for AI to perform some tasks in an occupation
Observed use
Share of tasks where AI tools are actually observed
Automation / augmentation
Whether a tool substitutes for work or assists a person
Employment outcome
Measured change in hiring, headcount, hours or wages
Evidence maturity
International comparisons and early evidence
Source traceability
Methods, dates and reports are accessible
Certainty of job loss
Whole-occupation loss remains uncertain
Need for updates
Tools, adoption and employment move rapidly
01 · TASK → OUTCOME
Start with the bundle of tasks, not the job title
Task exposure
Find units such as drafting, classification and summarisation.
Job redesign
People retain review, approval, face-to-face and field judgement.
Hiring and headcount
Productivity, demand and organisational choices shape staffing.
Wages and distribution
Technology alone does not decide who receives productivity gains.
02 · WORK MAP
Five work zones instead of one replacement list
Fast automation pressure
Data entry, routine documents, basic translation, simple support
Repetitive digital inputs with clear output formatsHigh augmentation potential
Coding, design, accounting review, legal research, medical documentation
Speed rises; verification and responsibility remain humanField hybrids
Maintenance, construction, logistics, quality inspection
Information can automate while physical settings varyHuman accountability
Safety, clinical decisions, hiring, education, care, disputes
Trust, context and liability are centralGrowing bridge roles
AI operations and audit, data stewardship, cyber security, workflow design
Connecting systems safely matters more than a model alone03 · KOREA
In Korea, exposure and inequality must be read together
The Bank of Korea estimates 51% of Korean jobs are highly affected by AI: 24% high-exposure/high-complementarity and 27% high-exposure/low-complementarity. These are task-based estimates, not observed dismissal rates.High AI impact
Bank of Korea 2025 estimate
High exposure · high complement
Potential productivity pairing
High exposure · low complement
Watch substitution and income risk
No overall staffing change
GenAI-using SMEs · OECD 2025
04 · ENTRY LADDER
The youth question is whether the first rung is narrowing
From July 2022 to July 2025, Korean youth jobs fell by 211,000; one analysis places 208,000 of the decline in high-AI-exposure industries. Demography and pre-existing trends also matter, so the full decline cannot be assigned causally to AI.Entry roles in clerical and ICT work show change early.
Automating junior tasks can remove the first stage of skill formation.
Redesign apprenticeships, review roles and training—not only job counts.
05 · FIELD CASE
Field maintenance changes tools before it loses people
In HVAC and equipment maintenance, AI can assist reports, fault-pattern analysis, inventory and scheduling. Disassembly, measurement, electrical and leak safety, exceptions and final accountability remain with the technician.AI assistance
Manual search · checklist · anomaly pattern · quote draft
Human measurement
Noise · vibration · temperature · pressure · site context
Accountable action
Isolation · repair · commissioning · customer explanation
06 · TIME HORIZONS
Keep observation, projection and scenario clocks separate
Observed now
Early signals in tool use, youth employment, adoption and job redesign
Employer projection
WEF projects 170m jobs created and 92m displaced across all macrotrends, net +78m. It is neither an AI-only effect nor a guaranteed outcome.
Structural scenario
Education, labour law, distribution and robotics can send hours and wages along very different paths.
07 · CLAIM AUDIT
Popular claims and what the evidence can support
AI will replace every job
Occupations bundle tasks and accountability. Current evidence supports broad transformation, not universal replacement.
UnsupportedA net 78 million jobs by 2030 is guaranteed
The WEF figure is an employer-survey projection across all macrotrends, not an AI-only forecast.
Conditional projectionHigh exposure means high unemployment
Exposure includes both automation and augmentation; outcomes depend on demand, firms, institutions and time.
Category errorPrompt engineer is the defining future job
The skill may be absorbed into domain roles rather than remain a standalone title.
WatchPhysical work is safe
Change may be slower, but scheduling, diagnosis, logistics and robotics alter parts of it.
Partly trueProductivity automatically raises wages and leisure
Distribution depends on bargaining, competition, institutions and ownership.
Not automatic08 · PREPARATION
Build portable capability instead of betting on one job
Problem framing
Decide what to automate and what to preserve
Verification
Check errors, evidence, safety and final decisions
Domain depth
Know the field well enough to judge AI output
Data judgement
Measurement, privacy, bias and records
Human work
Negotiation, explanation, care, teaching and conflict
Transition learning
Durable foundations and repeated retraining
09 · EVIDENCE LEDGER