86% Use AI at Work, Only 24% Feel Equipped: Firms Turn to Skills Mapping

Eighty-six percent of employees use AI tools at work, but only 24% feel fully equipped to use them effectively, according to Skillsoft research published in June 2026. That 62-point gap is pushing companies to scrap self-reported skills profiles and build assessment-based maps of AI proficiency across their technology teams — separating people who use AI from people who build with it.

The perception gap runs both ways. Seventy-seven percent of leaders believe they have set employees up for success, a 53-point divergence from what employees actually report. The Skillsoft survey, which covered 2,000 full-time employees, managers, and executives in North America, the UK, and Germany between March and April 2026, also found that only 11% of employees use formal skills assessments or benchmarks, 69% are only somewhat or not very clear on which skills matter, and just 28% strongly agree their job description matches their daily work. Only 16% received training before AI tools were rolled out. Note the caveat: the survey was commissioned by a company that sells skills management software.

"Organizations cannot afford to confuse AI adoption with AI readiness," says Ciara Harrington, Chief People Officer at Skillsoft. Her company argues the fix requires a "continuous view of workforce capability" — not a one-time audit.

What does Mexico's labor market show?

PwC's 2026 AI Jobs Barometer, Mexico edition, recorded about 68,000 job postings requiring AI skills in 2025 — 1.7% of all postings in the country. The split matters for anyone job-hunting. Positions for AI users grew by roughly 32,300 during the year, while AI developer positions increased by about 2,200, or 59.5% year over year. The technology, media, and telecommunications sector holds the largest share of developer roles at 11.3%.

"The labor market is evolving along two paths: companies are looking for people who can use AI in their daily work," says Claudia Zarco, Managing Director of Workforce Transformation at PwC Mexico. PwC also found that occupations highly exposed to AI recorded an average net skills change score of 2.74 between 2021 and 2025, versus 1.59 for less-exposed occupations — meaning AI-exposed roles are changing their skill demands faster.

Why defining the competencies is hard

The user-versus-developer split forms the first layer of most skills maps. Defining what sits inside each layer is the harder problem. According to the ETS 2026 Human Progress Report, 77% of Mexican human capital leaders find it difficult to identify which AI skills employers need, compared with 73% internationally. In the same survey, 83% of Mexican respondents said industry-specific AI competency standards would clarify what employers value.

Companies are building their maps through three methods:

• Task-based assessment. Codility's Skills Intelligence assigns real-world coding tasks inside a development environment, where employers can enable or disable an AI assistant per task. The platform logs every prompt and response and rates proficiency from novice to expert against its Engineering Skills Model 2.1, covering 210 skills and subskills across five categories. It measures assigned tasks only — it does not analyze everyday commits or repositories.
• Staged capability ladders. On Aug. 18, 2026, Pluralsight launched AI Ready, a managed program for engineering teams that use AI coding tools but haven't moved past basic autocomplete. It structures progress in three stages: AI-assisted coding, AI agent development, and multi-agent orchestration. Skill IQ assessments find gaps; a capstone requires engineers to build an end-to-end multi-agent workflow on a real codebase. Pluralsight keeps this separate from its AI Academy, which builds foundational fluency for technical and non-technical staff — mirroring the user-developer divide in PwC's data.
• External credentials. Here Mexican workers face specific constraints. Per the ETS report, 88% of Mexican respondents see certifications as essential to stay professionally relevant, but 65% struggle to obtain them, versus 48% globally. Only 37% have access to microcredential programs, though 87% want it.

What changes for individual practitioners

If you are an engineer or an AI user in Mexico, verification is now the currency. Access to tools "alone does not drive results," says Michael Ross, Chief Product Officer at Pluralsight. Self-reported profiles, résumés, and job descriptions remain the default skills data in many organizations, yet Skillsoft's numbers show they rarely match daily work. Assessment-based maps give employers a basis for staffing and upskilling decisions — particularly as Mexican AI roles expand beyond a small developer base toward a much larger population of AI users across sectors.

With no shared, industry-specific competency standards — which most Mexican human capital leaders say would help — each company currently defines its own taxonomy. That fragmentation means demonstrated, verifiable ability will likely keep gaining weight over declared skills as Mexico's AI hiring market matures.