Our Methodology

Data Sources

PlainWorkforce combines two official federal government datasets to provide a comprehensive view of occupational employment outlook and artificial intelligence workforce impact:

• Bureau of Labor Statistics Employment Projections 2024–2034: Official 10-year employment projections covering 832 occupations and 291 industries. Includes projected employment levels, growth rates, job change numbers, median annual wages, and typical education/training requirements. We also reference BLS OEWS (Occupational Employment and Wage Statistics) for wage cross-validation.
• O*NET Database 30.0 (US Department of Labor): The Occupational Information Network, which documents the tasks, skills, knowledge, abilities, and work activities required for each occupation. We use the Work Activities and Task statements to calculate AI exposure scores. Apprenticeship counts are drawn from DOL Office of Apprenticeship FY2024 data.

Employer AI-Exposure Profiles

Our Employers section profiles publicly-traded companies with a company-level AI displacement estimate. This section draws on a third source:

• SEC EDGAR company filings: employee counts, three-year headcount trend, ticker symbol, and SIC/NAICS classification come from each company's most recent 10-K XBRL filing. We currently track 1,965 public employers.

An employer's AI risk score is not computed from company-specific task data, it is estimated by mapping the employer's primary industry (SIC code) to BLS industry-occupation employment distributions, then applying the same O*NET task-level exposure scores used elsewhere on the site. This is a industry-average estimate, not a company-specific analysis of that employer's actual workforce composition.

AI Exposure Score Calculation

AI exposure scores (0–100%) represent the estimated proportion of an occupation's work content that is susceptible to current and near-term AI automation or augmentation. The methodology:

1. Task extraction: For each occupation, we retrieve all documented task statements from O*NET Database 30.0.
2. Task classification: Each task is evaluated against four AI automability criteria:
   • Routine data processing or structured information handling
   • Pattern recognition from standardized inputs
   • Rule-based or algorithmic decision making
   • Predictable physical actions in constrained environments
3. Weighted scoring: Task importance weights from O*NET are applied. The occupation's AI exposure score is the weighted proportion of its task content meeting automability criteria.

These scores reflect current and near-term AI capabilities as of the dataset vintage. They are analytical estimates, not predictions of employment outcomes.

Career Outlook Score (transparent derived index)

Each occupation page shows a 0–100 Career Outlook Score and an A–F grade. It is a transparent derived index computed by PlainWorkforce, not an official BLS, O*NET, or Department of Labor rating. Those sources publish the underlying wage, projection, and task data; PlainWorkforce combines them for comparison. It does not predict any individual's outcome.

• Inputs: the occupation's BLS median annual wage, its BLS Employment Projections 2024–2034 percent change, and its AI exposure score (see AI Exposure Score Calculation above).
• Dimensions and weights: Median wage 0.35, Projected growth 0.35, AI exposure (inverted) 0.30.
• Scoring each dimension: each value is converted to its percentile rank among every tracked occupation with that field on file: the share of occupations whose value is at or below it, rounded to a whole number. For AI exposure the percentile is inverted (100 minus the percentile), so lower exposure scores higher.
• Overall score: the weighted average of the dimension scores, rounded to a whole number. A dimension with no data is shown as N/A and its weight is dropped from the average instead of counting as zero.
• Grade thresholds: A+ 90 and above, A 85, A- 80, B+ 75, B 70, B- 65, C+ 60, C 55, C- 50, D 40, F below 40.

The score describes an occupation's standing across those three dimensions within this dataset, not a guarantee of future outcomes. A reader can reproduce a grade from the disclosed inputs, weights, and percentile method.

Employment Projections

Employment projections are taken directly from BLS as published, without modification. BLS projects employment levels for 2024 and 2034, the difference being the 10-year change. We also present the BLS-published median annual wage from the May 2024 OEWS survey as context.

Occupation Classification

Occupations use Standard Occupational Classification (SOC) codes. BLS projections and O*NET data are matched by SOC code. Where SOC codes differ between BLS and O*NET editions, we use the BLS projections coding as the primary key.

Important Caveats

Understanding the limitations of both employment projections and exposure scores is essential for responsible interpretation:

• Artificial intelligence exposure scores measure task content susceptibility, not the probability that any given worker will actually lose their position.
• Actual workforce effects depend on many factors including adoption speed, economic conditions, regulatory environment, and augmentation models where artificial intelligence assists workers rather than replacing them entirely.
• Bureau of Labor Statistics projections are inherently uncertain, they assume continuation of current economic trends and policies over the projection period.
• High exposure does not necessarily mean job loss; many high-exposure occupations are still projected to grow in employment due to other favorable economic factors and increasing demand.

Processing Pipeline

We combine BLS employment projections with O*NET task data through a multi-step process:

• Download BLS Employment Projections tables for the 2024–2034 projection cycle, extracting employment levels, growth rates, wages, and education requirements for each occupation
• Download O*NET Database 30.0 task statements and work activity importance ratings for each SOC-coded occupation
• Match BLS occupations to O*NET occupations by SOC code, handling code crosswalk differences between database editions
• Apply the AI automability classification to each O*NET task statement and compute weighted exposure scores
• Build industry-level views from BLS industry projections tables with growth rate and employment change data
• Compute sector and education-level aggregations for overview dashboards

Employment projections and wage data are presented exactly as published by BLS. AI exposure scores are our computed analytical metric based on O*NET task analysis.

Corpus placement (#N of M)

Occupation pages show where each SOC sits in the PlainWorkforce corpus, not just the occupation's own metrics. Placement uses two ladders that can diverge, so a single "#N of M" never pretends one metric is the whole story.

• Employment volume: rank among occupations with 2024 employment > 0, ordered by BLS employment_2024 (largest headcount = #1). Home health aides typically lead this ladder.
• AI exposure: rank among occupations with a scored O*NET task-automability estimate, ordered by ai_exposure_score (highest = #1). Bookkeeping and similar clerical roles can lead here while sitting mid-pack on headcount.

Ties break by occupation title (A-Z). Ranks are recomputed from the live occupations table (cached in-process). They are inventory placement inside this corpus, not a prediction that any worker will lose a job.

Not Affiliated

PlainWorkforce is not affiliated with the Bureau of Labor Statistics, the U.S. Department of Labor, or any government agency.

Data licensing and reuse (readiness)

PlainWorkforce does not sell data feeds or API keys today. The following assets are already published for citation and offline review, any commercial redistribution, embedding, or bulk extraction beyond the terms below requires operator approval.

• Occupation AI-exposure landscape CSV, occupation-ai-exposure-landscape.csv (also linked from /statistics). The underlying BLS/O*NET public-domain facts are free to use; PlainWorkforce's own AI-exposure scoring methodology, employer profiles, and editorial copy remain PlainWorkforce's copyrighted work per the Terms, Acceptable Use clause: personal, journalistic, academic, and non-commercial research use with attribution is permitted, systematic scraping or bulk redistribution requires prior written permission.
• llms.txt discovery file, /llms.txt lists indexable surfaces for AI crawlers; no authenticated API.
• Per-page CiteThis blocks, copy-ready citations on the homepage, methodology, statistics, and every major page; not a programmatic export.
• Affiliate posture (YMYL-adjacent): no affiliate links exist today. PlainWorkforce's AI-exposure scores inform real career and livelihood decisions, so any future affiliate unit would be barred from placement on or near a page for an occupation flagged high-risk (a direct conflict of interest, for example a resume or job-training service pitched next to a "your job is at risk" verdict), clearly labelled, and never mixed with a computed exposure score.
• Alternative ad networks: not activated; AdSense is the current monetization path. Mediavine or similar requires sustained real traffic first.