## GEN-AI: ARTIFICIAL INTELLIGENCE AND THE FUTURE OF WORK

## Source details

**Canonical URL:** [GEN-AI: ARTIFICIAL INTELLIGENCE AND THE FUTURE OF WORK](https://www.imf.org/-/media/files/oap/oap-home/2024/aisdnpptmarch14.pdf)

## Other formats

- [Markdown version](/-/media/files/oap/oap-home/2024/aisdnpptmarch14.pdf.md)
- [Structured JSON version](/-/media/files/oap/oap-home/2024/aisdnpptmarch14.pdf.json)

---

### Motivation and focus
- Artificial Intelligence (AI) is set to profoundly change the global economy.
- The Staff Discussion Note examines:
  - implications of AI adoption on jobs across AEs and EMDEs;
  - AI’s potential to displace and complement human labor;
  - potential effects of AI on inequality and productivity;
  - countries’ preparedness to adopt AI.

### AI exposure and complementarity: measurement and patterns
- Conceptual approach
  - Jobs are treated as bundles of tasks; some tasks can be performed by AI (exposure to AI by Felten, Raj, and Seamans, 2021;2023).
  - Shielding factors (social, ethical, physical context, and skill levels) reduce displacement risk (Index developed by Pizzinelli et al., 2023).
  - Complementarity potential derives from a combination of high AI exposure and high shielding.
- Examples
  - Judges: High AI exposure yet shielded by societal norms and laws—AI may complement their work.
  - Clerical Workers: High AI exposure with low shielding—higher displacement risk.
- Cross-country and demographic patterns
  - About forty percent of workers worldwide and sixty percent in AEs are in high-exposure occupations.
  - Employment shares by exposure and complementarity:
    - AEs: 27% high-complementarity; 33% low complementarity jobs;
    - EMs: 16% high-complementarity; 24% low complementarity jobs;
    - LICs: 8% high-complementarity; 18% low complementarity jobs.
  - Labor force composition (broad occupational groups) largely explains cross-country differences in exposure and complementarity.
  - Exposure is higher for women and for more educated workers, but is mitigated by a higher potential for complementarity with AI.
  - Exposure is spread along the labor income distribution but potential gains from AI are positively correlated with income.

### AI, productivity, and inequality: model-based analysis
- Model setup
  - Task-based model by Rockall, Pizzinelli and Tavares (2023) calibrated to the UK economy.
  - Model incorporates differences in labor productivity, asset holdings, AI exposure, and complementarity.
- Four critical channels of AI impact:
  1. Labor displacement: tasks shift from human labor to AI capital, reducing labor income.
  2. Complementarity: value added shifts to AI-complementary occupations, increasing labor demand there and reducing it elsewhere.
  3. Productivity gains: overall economic boost potentially offsets labor income losses.
  4. Capital income: AI adoption increases the return to capital, raising capital income.
- Scenario outcomes (model insights)
  - For all scenarios, the calibrated change in the capital share is the same: 5.5pp.
  - Scenario 1 — Low Complementarity:
    - Output increases by nearly 10%.
    - Change in total income by percentile shows distributional impacts (labor vs. capital decompositions).
  - Scenario 2 — High Complementarity:
    - Sectoral shift toward high-complementarity occupations.
    - Income increase is similar to Scenario 1; wage inequality rises.
  - Scenario 3 — High Complementarity and High Productivity:
    - Output surges by 16%.
    - Increase in total national income is largest and benefits all workers, although gains are larger for those at the top.
    - Aggregate measures: wage Gini and wealth Gini change; TFP and output rise.

### Potential for worker reallocation: historical evidence and life-cycle effects
- Transition patterns
  - Workers with college education have historically shown a greater ability to transition into occupations with high AI-complementarity potential.
  - One third of college-educated workers in “at risk” jobs are able to transition to jobs with high AI-complementarity potential.
  - Non-college-educated workers are predominantly found in low-AI-exposure jobs and are less inclined to move to high-complementarity positions when they switch from “at risk” jobs.
- Life-cycle implications
  - Younger workers:
    - Risk of missing stepping-stone jobs, making labor market entry more difficult.
    - AI may enable young college-educated workers to become experienced and productive more quickly.
  - Older workers:
    - May be less adaptable, face additional barriers to mobility, and have lower likelihood of reemployment after termination.
    - May have less incentive or fewer opportunities to learn new technologies.

### AI preparedness: cross-country readiness and policy prioritization
- AI Preparedness Index (AIPI)
  - Measures readiness across multiple strategic AI adoption areas, building on technology diffusion research.
  - Four themes: Digital infrastructure; Innovation and economic integration; Human capital and labor market policies; Regulation and ethics.
- Cross-country patterns
  - Higher-income economies, including AEs and some EMs, are generally better prepared than LICs to adopt AI.
  - Analysis covers 125 countries: 32 AEs, 56 EMs, and 37 LICs (plots use medians and group averages).
- Reform prioritization by development level
  - Foundational AI preparedness (digital infrastructure and human capital) is crucial for LICs and many EMs.
  - Second-generation preparedness (innovation and legal frameworks) is crucial for AEs (and some EMs) with already strong foundational preparedness and digital skills.
  - Correlations shown between ICT employment share and AIPI components suggest differing priorities:
    - Digital Infrastructure: Corr. = 0.6311 (AEs), Corr. = 0.4986 (EMs), Corr. = 0.3480 (LICs).
    - Human Capital & Labor Market Policies: Corr. = 0.3800 (AEs), Corr. = 0.3670 (EMs), Corr. = 0.4659 (LICs).
    - Innovation & Economic Integration: Corr. = 0.4111 (AEs), Corr. = 0.3995 (EMs), Corr. = 0.1152 (LICs).
    - Regulation & Ethics: Corr. = 0.4294 (AEs), Corr. = 0.4134 (EMs), Corr. = 0.2977 (LICs).

### Key findings and policy recommendations (summary)
- Key findings
  - Almost 40% percent of global employment is exposed to AI.
  - 60% of AE jobs are exposed to AI, mostly cognitive roles.
  - AI exposure: 40% in EMs, 26% in LICs.
  - AEs are generally at greater risk but also better poised to exploit AI benefits than EMDEs.
  - AI will impact income and wealth inequality; strong productivity gains could raise incomes for most workers, but distributional effects depend on complementarity and productivity outcomes.
  - Young, college-educated workers are better prepared to transition from jobs at risk of displacement to high-complementarity jobs; older workers may be more vulnerable.
- Policy recommendations
  - For AEs and better prepared EMs:
    - Focus on AI regulation and invest in AI innovation and integration.
  - For EMDEs:
    - Prioritize digital infrastructure and training to build foundational AI preparedness.
  - For all economies:
    - Strengthen social safety nets and retraining for AI-susceptible workers to ensure inclusivity.

---


_Source: https://www.imf.org/-/media/files/oap/oap-home/2024/aisdnpptmarch14.pdf_