AI Champions Program: Who Teaches AI? [2026]

An AI champions program starts with a question most L&D leads skip: who is already teaching? When AISA asked 90 people starting the AI fluency assessment, "Have you ever taught a colleague how to use an AI tool?", 71.1% said they had done so at least once. Another 27.8% said they had not but would like to. Only 1.1% said it was not their thing. The peer teachers you need for an AI champions program are almost certainly already on your payroll.

1. Yes, regularly35.6%32
2. A few times33.3%30
3. Once2.2%2
4. No, but I'd like to27.8%25
5. No, not my thing1.1%1

90 people answered, 16 September to 21 September 2026. Source: AISA research, aisa.to

Those numbers matter because formal training is not keeping up. Only 39% of people who use AI at work have had AI training from their company, according to the Microsoft and LinkedIn Work Trend Index (May 2024). And only 36% of employees feel adequately trained in AI use, per BCG's AI at Work 2025 report, which surveyed over 10,600 workers in 11 countries. The gap between what organisations provide and what people need is wide. Peer learning fills it.

This post breaks down the AISA research, explains what the numbers mean for anyone planning an AI champion role, and lays out a practical path to build a programme that actually spreads AI fluency through your teams.

How Many People Have Taught a Colleague to Use AI?

Most people have already shared AI knowledge with a colleague, and a large share do it regularly. Of the 90 people AISA asked, 35.6% said they teach colleagues regularly, 33.3% said they had done so a few times, and 2.2% said once. That adds up to 71.1% who have taught at least once.

The split between "regularly" and "a few times" is worth noting. The 35.6% who teach regularly are your strongest candidates for a formal AI champion role. They are not waiting for permission. They are already running informal demos, answering Slack questions, and showing teammates how to write better prompts.

The 33.3% who have helped a few times represent a second tier: people with enough skill and willingness to teach, but who may lack the time, confidence, or organisational backing to do it consistently. A structured programme gives them that backing.

What This Means for L&D Planning

If roughly seven out of ten people in your pipeline have already taught someone, the bottleneck is not willingness. It is structure. Most organisations treat AI training as a top-down initiative: buy a course library, assign modules, track completions. The data suggests the real learning is already happening sideways, colleague to colleague, without any formal support.

The practical move is to find those peer teachers, give them a title, a time budget, and a clear scope, and let them do what they are already doing, but better.

Who Are the Informal AI Teachers at Work?

Informal AI teachers are the people who show a colleague how to draft a prompt, recommend a tool for a specific task, or troubleshoot an AI output over someone's shoulder. They are not necessarily the most senior people on the team. They are the ones others already go to with questions.

Spotting Them Without a Survey

You can identify informal teachers by looking at behaviour signals:

• Who shares AI tips in team channels? Check Slack or Teams for people who post prompts, tool recommendations, or workflow shortcuts.
• Who gets tagged in AI-related questions? The person others mention when they are stuck is your informal champion.
• Who experiments visibly? People who demo new tools in meetings or share before-and-after examples of AI-assisted work are signalling both skill and willingness.

Skill Level Matters Less Than You Think

A common mistake is assuming AI champions need to be technical experts. They do not. The most effective peer teachers are often people who understand the work context deeply and can translate AI capabilities into practical steps for their function. A product manager who shows designers how to use AI for user feedback synthesis is more useful than a machine learning engineer explaining transformer architecture to a marketing team.

What matters is the combination of competence, credibility within their team, and the willingness to teach. The AISA data shows willingness is widespread. Competence is what you need to measure, and a team AI assessment gives you a baseline for each person so you can match champions to the right scope.

How Many People Want to Teach AI but Have Not Yet?

27.8% of the 90 people AISA asked said they had not taught a colleague but would like to. That is 25 people who have the motivation but have not found the opportunity, the confidence, or the context to start.

This group is your untapped programme capacity. They are not reluctant. They are ready. The barrier is almost certainly structural: no one asked them, there is no format for it, or they are unsure whether their knowledge is "enough."

Turning Willingness into Action

For this group, the intervention is lightweight:

• Give them a first assignment. Ask them to run a 15-minute "show and tell" in their next team meeting. A single, low-stakes session converts intention into habit.
• Pair them with an experienced champion. If you already have people in the 35.6% who teach regularly, pair a willing-but-inactive person with them for a month. Observation lowers the barrier.
• Remove the expertise expectation. Make it clear that an AI champion does not need to know everything. They need to know enough to help one colleague with one task. Task decomposition is a useful frame here: break "teach AI" into small, specific actions like "show how to summarise meeting notes" or "help draft a first prompt for a recurring report."

Combined with the 71.1% who have already taught, you are looking at 98.9% of people who have either taught or want to. The 1.1% who said "not my thing" is a single person out of 90. Resistance to peer teaching is, in this data, essentially absent.

How Do You Choose People for an AI Champions Programme?

Selection should combine demonstrated behaviour, measured skill, and functional coverage. Picking only volunteers risks clustering champions in one department. Picking only high performers risks choosing people who are too busy to teach.

A Practical Selection Framework

Measuring Skill Before You Assign the Role

The AISA research tells you who is willing. It does not tell you who is skilled. Willingness without competence produces champions who spread bad habits: hallucination risks ignored, prompts that work by accident, tools chosen for novelty rather than fit.

A structured assessment gives you the skill data. AISA's published rubric covers prompting, critical thinking, technical understanding, workflow application, and safety. Champions should score at least Competent across these dimensions. If someone is strong on prompting but weak on safety, you know what to train before you deploy them.

This is where a team AI assessment pays for itself. Run it before you select champions, and you get a skill map of your entire team. You see who is ready now, who needs development first, and where the functional gaps are.

How Many Champions Do You Need?

A common ratio is one champion per 15-25 people. Smaller teams might need one per team regardless of size. The goal is proximity: every person should have a champion they interact with regularly, not someone three floors away in a different department.

What Does an AI Champion Actually Do?

An AI champion is a peer who helps colleagues adopt AI tools and practices within their day-to-day work. They are not trainers delivering courses. They are not IT support. They are practitioners who happen to be a step ahead and are willing to bring others along.

Core Responsibilities

1. Demonstrate practical use cases. Champions show, not tell. They run short demos of how they use AI for real tasks: drafting emails, analysing data, generating first drafts, reviewing code. The emphasis is on tasks the audience already does, not hypothetical future workflows.

2. Answer questions in context. When a colleague is stuck on a prompt or unsure which tool to use, the champion is the first point of contact. This is where most AI learning happens, in the moment, on a real task, with someone who understands the work.

3. Surface and share patterns. Champions notice what works across their team and codify it. A prompt that saves 30 minutes on weekly reporting becomes a shared template. A workflow that combines two tools becomes a documented process. This is how AI peer learning scales beyond one-to-one conversations.

4. Flag risks and reinforce good practice. Champions are also the front line for responsible AI use. They remind colleagues to check outputs, avoid sharing sensitive data with public tools, and follow the organisation's AI policy. If your company has an AI policy, champions are the people who make it real on the ground.

5. Feed insights back to L&D. Champions see what people struggle with, what tools they ignore, and what questions keep coming up. That feedback loop is more valuable than any survey. It tells L&D where to invest next.

What Champions Should Not Do

Champions are not responsible for tool procurement, security reviews, or enterprise AI strategy. They are not a substitute for formal training on high-stakes applications. And they should not be expected to support every person in the organisation. Scope the role clearly, or your best champions will burn out within a quarter.

How Do You Spread AI Skills Through Peers?

Peer learning works when it is structured enough to be consistent but informal enough to feel natural. The AISA data shows that 35.6% of people already teach regularly without any structure at all. Adding a light framework amplifies what is already happening.

Formats That Work

Show-and-tell sessions. A 15-minute slot in an existing team meeting where one person demonstrates an AI workflow. No slides. Just a screen share and a real task. These are low-effort to organise and high-impact because they are grounded in the team's actual work.

Prompt libraries. Champions curate and share prompts that work for their function. A shared document or channel where people can grab, adapt, and contribute prompts turns individual knowledge into team knowledge. This connects directly to building prompt templates that the whole team can use.

Office hours. A weekly 30-minute window where the champion is available for questions. This works well for teams that are distributed or asynchronous. It gives people a known time to bring problems without feeling like they are interrupting.

Pair sessions. A champion sits with a colleague for 20 minutes and works through a task together using AI. This is the highest-impact format per minute spent, because the learning is immediate, contextual, and hands-on.

Measuring Whether It Is Working

Peer learning is hard to measure with completion rates. Instead, track:

• Tool adoption rates across the team over time
• AI fluency scores before and after the programme (reassess quarterly)
• Number of use cases documented by champions
• Qualitative feedback from team members on confidence and frequency of AI use

The combination of a baseline team AI assessment and a follow-up assessment three to six months later gives you the clearest signal. If scores move, the programme is working. If they do not, you know to adjust.

For a broader view of how AI upskilling connects to measurement, see AI Upskilling and Measurement in 2027.

Sustaining Momentum

Champion programmes lose steam when champions feel unsupported. Three things keep them going:

1. Protected time. If champion work competes with their day job, the day job wins. Agree a time budget with their manager.
2. Recognition. This does not have to be financial. A mention in an all-hands, a badge in the internal directory, or a line in their performance review all signal that the organisation values what they do.
3. Their own development. Champions need to keep learning too. Give them access to advanced resources, connect them with champions in other functions, and let them attend external AI events. People who wonder whether they are keeping up are more motivated when they have a clear development path.

The formal training gap is real. With only 36% of employees feeling adequately trained in AI use (BCG, June 2025), and only 39% of AI users having received company training (Microsoft, May 2024), peer learning is not a nice-to-have. It is the mechanism that fills the gap between what organisations provide and what people need to be productive with AI.

If you want to know where your team stands before you launch a programme, start with a measured baseline. A team AI assessment gives you the skill map, and the champions programme gives you the path forward.

About this research

This finding comes from AISA's own research. AISA asks a short research question of people as they start the AISA AI fluency assessment, one question at a time. The question in the chart above was asked from 16 September to 21 September 2026, worded exactly as shown and with the answers in the order shown, and 90 people answered it. Each person counts once: when someone answered more than once, their latest answer is the one counted.

More findings from this research are collected in AISA research and data.

How to cite this research: AISA (2026). "AI Champions Program: Who Teaches AI? [2026]". AISA research. https://aisa.to/blog/ai-champions-who-teaches-ai-at-work

---

Related reading: Is Everyone Using AI at Work? [2026 Survey], see how widespread AI use actually is across roles.

Related reading: AI Upskilling and Measurement in 2027 [Strategy], a framework for connecting upskilling to measurable outcomes.

Related reading: Am I Falling Behind on AI? [2026 Data], what people worry about and what the data actually shows.

Frequently Asked Questions

What is an AI champion?

An AI champion is a peer within a team or function who helps colleagues adopt AI tools and practices in their daily work. They are not formal trainers or IT support. They are practitioners who are a step ahead in AI fluency and willing to share what they know through demos, answering questions, and documenting useful workflows.

How do you start an AI champions programme?

Start by identifying who is already teaching informally. Look for people who share AI tips in team channels, get tagged in AI questions, or demo tools in meetings. Measure their AI skill level with a structured assessment to confirm they are ready, then give them a clear scope, a time budget, and a first assignment like a 15-minute show-and-tell session.

How do you encourage employees to use AI?

The most effective way to encourage AI adoption is through peer learning. When a trusted colleague shows how AI helps with a task the team already does, adoption follows naturally. Combine this with protected time for experimentation, shared prompt libraries, and regular reassessment so people can see their own progress. Formal training alone reaches only a fraction of the workforce.