Making People Use AI — Training and Onboarding

A month after rollout, only 20% of the team uses the AI tool daily, because learning costs feel bigger than the payoff. How to build champion users and reduce resistance.

AXAI TransformationTrainingOrganizational AdoptionConsulting

No matter how good the tool, it's useless without users. AI adoption's real challenge isn't technology — it's people.

The Tool Exists But Nobody Uses It

One month after AI deployment. Logs show only 20% use it daily. The other 80% tried once or twice, then went back to old methods.

This is normal. Most AI tool deployments face this reality.

Three reasons: learning cost feels higher than benefit, fear of AI-caused mistakes, and the sheer force of habit.

Start With Champion Users

Don't train everyone at once. Create champion users first — people who use AI early, discover good practices, and naturally become the team's go-to.

One to two per team is enough. They receive deeper training first, then spread to their teammates. People move more easily from "my colleague says it's really convenient" than from IT department mandates. Champion advocacy, backed by organizational support, is what lifts real usage.

Moderna took this route. According to OpenAI's Moderna case study (April 2024), Moderna ran an internal AI prompt contest to identify its top 100 AI power users and organized them as a cohort of internal Generative AI Champions. Within two months of rolling out ChatGPT Enterprise, the company had 750 GPTs, and 40% of weekly active users had created GPTs of their own.

Korea's public sector chose the same structure. In September 2025, the Ministry of the Interior and Safety announced (Korean) a plan to train 20,000 internal "AI Champions" (2% of all public-sector staff) by 2030 to lead AI transformation in each agency. The training is hands-on, and certification is granted only after passing a practical assignment assessment. Rather than turning every employee into an expert at once, it plants anchor talent on the front line.

Training Design Principles

Under 30 minutes. Two-hour AI training means 90 minutes forgotten.

Use our actual work data. Generic examples get "interesting" and nothing more.

Provide immediately usable prompt templates. 3-5 templates that participants take home — these matter more than the training itself.

Say failure is okay. "AI might give wrong answers. That's AI's limitation, not your mistake." This one sentence dramatically lowers the psychological barrier.

Handling Resistance

"Will my job disappear?" → "AI takes repetitive tasks, not your role." Show new roles concretely.

"The old way is easier." → Don't force. Offer comparison: "Try half old way, half AI this week." Self-chosen adoption lasts longer.

"No time to learn." → Embed AI into existing workflow, not on top of it. Details in the next article.

Onboarding Timeline

Day 1: 30-min training + prompt templates. "Use it just once today." Week 1: Champion shares tips in team channel. Week 2: Check usage, ask non-users why 1:1, remove barriers. Month 1: Share data. "AI users saved an average of 3 hours/week." Numbers move the rest.

Training Isn't One-Time

Monthly 15-minute "AI tips" sessions. Champion users share one or two useful discoveries. Low effort, high impact.

Training has to accumulate for usage to grow. In BCG's AI at Work 2025 survey (over 10,600 workers in 11 countries), only 36% of employees felt adequately trained in AI, and those who received five or more hours of training, especially in person and with coaching, were significantly more likely to become regular users. A 30-minute kickoff is only the start; short follow-up sessions need to keep coming.

AI's success or failure depends not on technology, but on whether people actually use it.