Research question
How often did a purposive sample of 100 current U.S. management vacancies from companies in the 2026 Fortune 100 contain an explicit, role-specific artificial-intelligence signal, and how did that signal vary by management function and employer?
Abstract
MTF Institute coded 100 distinct U.S. management vacancies from eight employers in the 2026 Fortune 100. Forty vacancies contained an explicit AI signal in the role title, responsibilities or qualifications after manual removal of career-site navigation, related-job lists and employer-branding modules. Twenty involved building or deploying AI/ML, 16 involved using or enabling AI, three involved governing or evaluating AI, and one mentioned AI without making the role relationship sufficiently specific. Explicit AI exposure was concentrated in engineering/data/technology (12 of 16) and product management (6 of 9), while none of the 15 operations/facilities/construction vacancies or eight health/care-delivery vacancies in this sample disclosed an explicit AI signal. The result is a dated, purposive snapshot rather than a Fortune 100 population estimate.
Report number: MTF-RR-2026-08-26-01
Publication date: 26 August 2026
Author: MTF Institute Editorial Team
Institution: MTF Institute
DOI: 10.5281/zenodo.22104696
Scope and method
The sampling frame was the 2026 Fortune 100. The study used a frozen purposive corpus of 100 unique public U.S. vacancies captured from official employer career pages: 99 on 10 August 2026 and one replacement U.S. vacancy on 24 August 2026. The sample covers eight employers and ten functional families. It intentionally maximises functional and title diversity within accessible current management postings; it is not a random or statistically representative sample of all Fortune 100 vacancies.
A valid observation required:
- a distinct public vacancy URL on an official employer career site;
- a management, director, programme, product, professional-leadership or equivalent responsibility level;
- a U.S. location;
- sufficient title, responsibility or qualification text to code the role; and
- inclusion in the frozen 100-record source inventory.
Duplicates, non-U.S. postings, inaccessible records and page elements unrelated to the vacancy were excluded. One Germany posting in the original capture was replaced with a current U.S. Amazon Area Manager vacancy so the analytical frame remained U.S.-only.
Coding framework
The codebook separated role-specific AI exposure from adjacent digital language.
| Class | Coding rule |
|---|---|
| Build or deploy AI/ML | The role builds, deploys, scales, productises or technically operates AI, ML, LLM or generative-AI capability |
| Use or enable AI | The role applies AI tools, enables adoption, sells/partners around AI or incorporates AI into a business workflow |
| Govern or evaluate AI | The role addresses AI risk, privacy, compliance, controls, evaluation, ethics or oversight |
| Explicit AI signal; relation unspecified | AI is role-specific, but the vacancy does not support a stronger relationship classification |
| Adjacent without explicit AI | Data/analytics or automation/digital-transformation language appears without an explicit AI term |
| No disclosed signal | No role-specific explicit AI or adjacent signal was found |
Automated pattern matching generated a review queue. Every apparent explicit-AI positive was then checked against the vacancy's title, responsibilities and qualifications. Matches in job recommendations, related-job lists, benefits, testimonials, navigation and generic career-site promotion were removed. The archived coding file preserves the decision, evidence excerpt and source URL for all 100 observations.
Main result: 40 of 100 vacancies disclosed explicit AI exposure
| Exposure class | Vacancies | Share of sample |
|---|---|---|
| Build or deploy AI/ML | 20 | 20% |
| Use or enable AI | 16 | 16% |
| Govern or evaluate AI | 3 | 3% |
| Explicit AI signal; relation unspecified | 1 | 1% |
| Adjacent without explicit AI | 4 | 4% |
| No explicit AI or adjacent signal | 56 | 56% |
| Total | 100 | 100% |
The distinction matters. “Forty percent mention AI” does not mean 40% are AI-engineering roles. Half of the explicit positives - 20 vacancies - involved building or deploying AI/ML. Sixteen connected AI to adoption, workflow, partnership, sales or productivity. Only three made governance or evaluation the dominant relationship.
Functional concentration was stronger than the overall average
| Functional family | Vacancies | Explicit AI | Explicit-AI share |
|---|---|---|---|
| Engineering, data and technology | 16 | 12 | 75.0% |
| Product management | 9 | 6 | 66.7% |
| Design, content and experience | 6 | 3 | 50.0% |
| Legal, risk and compliance | 10 | 5 | 50.0% |
| Programme, project and delivery | 12 | 6 | 50.0% |
| Commercial, marketing and partnerships | 12 | 5 | 41.7% |
| Finance and banking | 7 | 2 | 28.6% |
| General and organisational leadership | 5 | 1 | 20.0% |
| Operations, facilities and construction | 15 | 0 | 0.0% |
| Health and care delivery | 8 | 0 | 0.0% |
The zeros should not be read as proof that those functions do not use AI. They mean the sampled vacancies did not explicitly disclose AI in the role-specific text. This is a disclosure study, not an observation of work performed after hiring.
Employer variation shows the effect of the purposive frame
Explicit AI appeared in 13 of 15 sampled Alphabet (Google) vacancies and eight of 12 Microsoft vacancies, but in none of five CVS Health vacancies and one of 13 UnitedHealth Group vacancies. Apple recorded seven of 18, JPMorgan Chase five of 12, Amazon four of 13 and McKesson two of 12.
These differences combine real portfolio effects with sample composition. The Google and Microsoft sets contained more technology and AI-facing leadership roles; the healthcare samples contained more operational and care-delivery management. The employer percentages therefore describe this corpus only and should not be used as a company-wide AI-readiness ranking.
What the snapshot suggests
1. AI capability is becoming a role design variable, not only a specialist title
Explicit signals appeared across product, programme, commercial, legal and finance roles as well as engineering. Managers increasingly need to understand whether they are expected to build AI, enable its use or govern its consequences.
2. Governance disclosure trails build and adoption disclosure
Only three vacancies placed governance or evaluation at the centre of the AI relationship. Some build and use roles also mentioned privacy, security or responsible practice, but the dominant vacancy signal was still delivery or enablement. Candidates should not assume that governance ownership is explicit merely because AI appears in the description.
3. Absence of the term is not evidence of absence of change
Fifty-six vacancies disclosed neither AI nor the narrow adjacent indicators used here. Their work may still be affected by automated tools, enterprise platforms or later operating-model changes. The correct conclusion is that the vacancy did not make that exposure visible.
Practical application: the SIGNAL-4 vacancy test
Students and career professionals can apply four steps to any vacancy.
- Scan the verbs. Highlight build, deploy, evaluate, govern, use, enable, automate and optimise. Ignore generic employer branding.
- Identify the relationship. Is the role expected to build AI, use it, enable others or govern it? Do not treat all mentions as the same skill requirement.
- Generate proof. Prepare one relevant evidence artifact: an evaluation scorecard, controlled workflow, adoption plan, model-risk memo, product experiment or value case.
- Ask the boundary questions. In an interview, ask which AI decisions the role owns, what data and tools are approved, how output is reviewed, which risks require escalation and how success is measured.
The study's practical implication is not “put AI on every résumé.” It is to match evidence to the disclosed relationship. A product leader needs different proof from a compliance director or operations manager.
MTF Institute's Executive Certificate in AI, Digital Transformation & Platform Strategy helps managers connect AI opportunity, platform choices, governance and change leadership. The SIGNAL-4 test can be used to select the parts of that learning most relevant to a target role.
Limitations
The sample is purposive, covers eight of the 100 ranked companies and is uneven across employers and functions. Public vacancy pages can change or disappear after capture. Coding measures disclosure in text, not actual tool use, investment, organisational maturity or employee performance. The keyword-assisted review may miss indirect language that describes AI work without using an explicit term. Percentages are descriptive for the frozen corpus and are not population estimates for the Fortune 100, U.S. management employment or any individual company.
Data and reproducibility
The Zenodo record contains a searchable PDF, the 100-row coding CSV and a supporting workbook with the source inventory, codebook, formula-driven summaries and quality-control checks. Each row includes the employer, title, function, source URL, capture date, exposure class and a bounded evidence excerpt. Source pages remain subject to employer availability and terms.