The AI Skills Gap on IT Teams: Measure It, Then Close It

Every IT leader I talk with agrees their team has an AI skills gap. Very few can tell me where it is. Is it the service desk, who’ll field users’ AI questions? The infrastructure team, who’ll be asked to connect AI to business systems? Security, who’ll be asked whether a tool is safe? Or is it a general sense that “we’re behind,” without anything specific enough to act on?

A vague skills gap produces vague responses: a lunch-and-learn, a subscription to an online course library that few people open, or a plan to “hire someone who knows AI.” A measured skills gap produces a plan. This post covers how to measure it by role, and the four levers that close it.

Why the gap is wider on IT teams than it looks

Most IT people have tried AI tools. That creates an impression of competence that doesn’t survive a closer look. Using a chatbot to rewrite an email is a very different skill from reviewing an AI-generated script before it runs in production, securing an AI integration, or explaining to a department head why an assistant gave a wrong answer.

IT also carries a double load. Like everyone else, IT staff need to use AI well in their own work. Unlike everyone else, they’ll be asked to support, secure, and build AI for the rest of the organization. The skills gap that matters is the second one, and it’s the one teams measure least.

The four skills questions in the assessment

The skills dimension of the AI Readiness assessment asks four questions. Each has its own guide in this series.

1. Hands-on skill: how hands-on is your IT team with AI tools? Most have tried; few are trained. See from “everyone has tried ChatGPT” to a trained team.
2. Certifications: how many IT staff hold current cloud or AI certifications? Coverage of key roles matters more than the count. See which AI certification your IT team should get first.
3. Builders: does anyone on staff build with AI? One named builder changes what’s possible. See you need one AI builder on staff.
4. Investment: how much time and budget go to AI upskilling? Case-by-case approval quietly signals that learning is optional. See the case for a standing AI training budget.

The pattern I see most often

The skills profile I see most at mid-size IT teams is high enthusiasm and low structure. Most people have tried AI tools, but nobody has had formal training. There are a few cloud certifications and nothing AI-specific. One person builds things on their own time, unofficially. And there’s no budget line for any of it, so every course or exam is a separate request. That profile is encouraging, because the interest is already there. What’s missing is organization, and organization is cheaper to add than enthusiasm.

The asset: an AI skills matrix for your IT team

This is how you turn “we’re behind” into something you can act on. Build a table with people down the side and skills across the top, grouped by role.

The skills to include

For everyone in IT:

• Writing effective prompts for their own work.
• Knowing what data never goes into an AI tool.
• Verifying AI output before relying on it.
• Explaining the approved tools and policy to a user.

For specific roles:

• Service desk: supporting users on the approved assistant; using AI for ticket summaries and knowledge articles.
• Infrastructure and cloud: reviewing AI-generated scripts; configuring cloud AI services; setting up budgets and logging.
• Security: AI-specific threats such as prompt injection; DLP for AI; reviewing AI vendors; reading AI audit logs.
• Builders: calling models through APIs; building automations and agents; testing AI output; version-controlling prompts and configurations.

The rating scale

• 0, none: hasn’t done it.
• 1, aware: understands it, hasn’t practiced.
• 2, practiced: does it competently in real work.
• 3, can teach: could show a colleague how.

How to fill it in

Have each person rate themselves, then have their manager adjust based on observed work. Self-ratings on AI tend to run high for skills people have tried once, so add one rule: a rating of 2 or 3 needs an example from real work. The whole exercise takes about an hour per team.

Reading the result

Look for three things. Skills where the whole team is at 0 or 1: those are training priorities. Critical skills held by only one person at 2 or 3: those are single points of failure. And roles where nobody reaches 2: those are the gaps that will slow down your AI plans first.

Four levers to close the gap

1. Structured training closes broad gaps, especially the “everyone” skills. A six-week, role-based plan with protected time moves most of a team from 1 to 2.
2. Certifications add depth and an outside benchmark for key roles. They work best when tied to a real project. The ranked guide to which AI certification to get helps you choose.
3. A named builder closes the building gap, which training alone rarely does. One person with protected time and a real project usually gets further than a whole team with a course.
4. A standing budget keeps the other three running. Without it, each lever depends on individual requests being approved.

A decision rule: grow, partner, or hire

Grow skills internally for roles that need knowledge of your environment. Partner for one-off specialist work. Hire only when you need a sustained capability you can’t grow within a year.

Most AI skills IT teams need are best grown, because they depend on knowing your systems, your data, and your people. A consultant can build a specialized solution faster, but the knowledge leaves with them. Hiring makes sense when you need a lasting capability, such as a small team building AI solutions, and you don’t have anyone who can grow into it in a reasonable time.

A 90-day plan to close the most important gaps

Days 1 to 15: build the skills matrix and identify the three biggest gaps.

Days 16 to 60: run the six-week training plan for the team, with role tracks aimed at the gaps you found. Name your first builder and start their 90-day plan.

Days 61 to 90: start the first certifications for key roles, submit the standing training budget, and rebuild the matrix to measure the change.

What closing the skills gap is not

It’s not a course library subscription. Access to courses doesn’t change skills; protected time and applied practice do.

It’s not one hire. An AI specialist without a trained team around them becomes a bottleneck.

It’s not a one-time push. AI tools change quickly. Rebuild the matrix twice a year.

How skills fit with the other five dimensions

Skills are one of six dimensions in the AI Readiness framework. They enable most of the others: someone has to configure the security controls, build the integrations, and run the governance process. For the whole framework, see the 6-dimension AI readiness framework, explained.

Where does your team actually stand?

The free AI Readiness Score includes three skills questions alongside the other five dimensions. It’s 10 questions and gives you a score in a few minutes.

Get your free AI Readiness Score →

Want to see what the full assessment covers first? Flip through a complete 38-page sample report.

Related guides

• From “Everyone Has Tried ChatGPT” to a Trained Team
• Which AI Certification Should Your IT Team Get First?
• You Need One AI Builder on Staff. Here’s How to Grow One
• The Case for a Standing AI Training Budget
• The 6-Dimension AI Readiness Framework, Explained