The AI-usage requirement
Two instruments bracket the AI-usage requirement between 1.4% and 8.1% of 998,166 live postings in 2026-07, measured 2026-08-11.
Updated
https://olive.is/answers/benchmarks/ai-usage-requirement/ · Edition 2026Q3| Month | Extracted skills (floor) | Raw text scan (ceiling) | Base | Status |
|---|---|---|---|---|
| 2026-02 | 0.5% | 3.8% | 445,637 | charted |
| 2026-03 | 0.5% | 3.2% | 2,559,476 | charted |
| 2026-04 | 0.7% | 4.5% | 993,788 | charted |
| 2026-05 | 0.8% | 6.8% | 1,429,361 | charted |
| 2026-06 | 1.1% | 8.9% | 1,023,700 | charted |
| 2026-07 | 1.4% | 8.1% | 998,166 | charted |
| 2026-08 | not charted: partial month, 45,683 postings |
Olive. The AI-usage requirement, edition 2026Q3. Measured 2026-08-11. https://olive.is/answers/benchmarks/ai-usage-requirement/ Accessed __.
https://olive.is/answers/benchmarks/ai-usage-requirement/ · Edition 2026Q3| Name | Rate | Base |
|---|---|---|
| Financial Services | 7.3% | 60,833 |
| Technology | 6.0% | 459,137 |
| Consulting | 5.0% | 97,173 |
| Information Tech. | 3.3% | 85,374 |
| Education | 1.6% | 64,159 |
| Banking/Finance | 1.5% | 47,214 |
| Engineering | 1.1% | 64,155 |
| Government | 1.0% | 100,312 |
| Healthcare | 0.7% | 496,911 |
| Manufacturing | 0.5% | 132,857 |
| Retail | 0.1% | 428,784 |
Olive. Concentration by industry, edition 2026Q3. Measured 2026-08-11. https://olive.is/answers/benchmarks/ai-usage-requirement/ Accessed __.
https://olive.is/answers/benchmarks/ai-usage-requirement/ · Edition 2026Q3| Name | AI-usage postings | All live postings | n | Base |
|---|---|---|---|---|
| Under 1 year | 0.8% | 4.8% | 236 | 28,735 |
| 1 to 3 years | 8.4% | 27.2% | 2,413 | 28,735 |
| 3 to 5 years | 17.9% | 20.2% | 5,140 | 28,735 |
| 5 to 8 years | 42.5% | 27.9% | 12,211 | 28,735 |
| 8 or more years | 30.4% | 19.9% | 8,735 | 28,735 |
Olive. Seniority bands, edition 2026Q3. Measured 2026-08-11. https://olive.is/answers/benchmarks/ai-usage-requirement/ Accessed __.
Checked
Unmeasured. No verification pass has run against this edition, so no checked-on date exists to print.
Source
Production Elasticsearch, index of live job postings
Extract
judgment-gate/data/index/1.json
Reproduce
One row per posting in the index snapshot, counted by whether the posting text states an AI-use requirement. The snapshot the count was taken from is named in `extract`.
23 rows. Every published, withheld and excluded cell. Creative Commons Attribution 4.0 International.
What this cannot tell you
- Two instruments bracket the level and neither fixes it. The floor depends on the skill extractor having caught the term and the ceiling counts a company describing itself as AI-powered, which requires nothing of a candidate.
- The industry figure carries the extracted instrument alone, so each of the eleven rates is a floor and none of them is comparable to the bracket in the climb.
- The eleven industries are a head list rather than a census, so they publish no total and the eleven bases do not sum to the corpus.
- The seniority figure compares two populations on the minYoE field, which is documented as fabricated upward. Read the gap between the bands, never the level of either.
- The years-of-experience instrument that bands the seniority figure is not one of the two AI-usage instruments listed above. Each seniority cell names its own term list.
- Monthly bases are not comparable in size: 2026-03 carries 2,559,476 postings against 445,637 in 2026-02, so the six shares are not a smoothed series.
- The 2026-08 row is excluded as a partial month at 45,683 postings. It ships in the CSV with that reason rather than being dropped.
- No cell is withheld at this edition. judgment-gate/data/index/1.json reports 30 cells considered, 30 published and 0 dropped.
- checkedOn is null. No verification pass has run since the values were built on 2026-08-25.
Claims this data does not support
- No point estimate for the requirement rate. The bracket stays a bracket until 300 postings per arm are hand-adjudicated.
- No corpus-level denominator on a monthly figure. The 2,151,213 live postings measured 2026-08-11 are neither a charted cell's base nor the charted period's sum.
- No percentile, no ranking, no cohort comparison, and no number standing for a person.
- No distribution of candidate performance across D1 to D6. The Index records assessment aggregates as notRun.
- No chart from the AEO panel. Every rate in olive-stories/data/panel/2026-08-25.json is null on n of 0, and a zero there would make every later change look like progress.
- No claim about the market. The corpus is a census of one index, and a posting the index never ingested is outside every figure here.
Place this chart
<!-- Olive embed. Licence: https://olive.is/licence/embed/1-0 -->
<iframe src="https://olive.is/embed/v1/chart/ai-usage-requirement"
width="640" height="1155" loading="lazy"
style="border:0;width:100%;max-width:640px;aspect-ratio:128/231"
sandbox="allow-popups allow-popups-to-escape-sandbox"
referrerpolicy="strict-origin-when-cross-origin"
title="Share of live job postings stating an AI-usage requirement"></iframe>
<p>Source: <a href="https://olive.is/answers/benchmarks/ai-usage-requirement/" rel="nofollow">Olive</a></p>
The attribution link sits outside the frame, so it survives a host that refuses framing. It carries rel="nofollow" and passes no authority, on purpose. Embed licence.