When I first introduced AI tools to my L&D team at the bank, I made every mistake in the book. I sent a company-wide email with a link to ChatGPT and said “start using this.” The result? About 15% of staff tried it once, got confused, and never touched it again. Another 30% were convinced their jobs were about to disappear. The rest ignored it entirely.

That experience taught me something important: training people on AI isn’t really about the technology. It’s about psychology, change management, and meeting people where they are.

Start with the “Why” That Matters to Them

Most AI rollouts fail because they lead with productivity gains and efficiency metrics. Nobody cares about making the company more efficient if they think they’re training their replacement.

Instead, start with individual benefits. What does this tool do for them?

• “You can draft that monthly report in 20 minutes instead of two hours”
• “That email you’ve been putting off? Let’s get a first draft in 30 seconds”
• “Stuck on how to phrase that difficult conversation? Here’s a sounding board”

The productivity gains will come. But first, you need buy-in.

The Three-Session Framework

After years of trial and error, I’ve landed on a three-session approach that actually works.

Session One: Demystification (90 minutes)

This session has one goal: make AI feel approachable. No productivity expectations, no metrics, no “you should be using this daily.”

Cover these elements:

• What ChatGPT actually is (and isn’t)
• Live demonstration of basic prompts
• Everyone writes their first prompt together
• Open Q&A about concerns

The most important part happens in the Q&A. Let people voice their fears. Don’t dismiss them. “Will this take my job?” is a legitimate question that deserves a thoughtful answer, not corporate spin.

Session Two: Practical Application (2 hours)

Now we get specific. Bring real work examples:

• Draft a tricky email to a difficult stakeholder
• Summarise a long document
• Create a first draft of a presentation outline
• Generate questions for an upcoming meeting

Everyone should leave with at least one workflow they’ll actually use. This is where the light bulbs start going on.

Session Three: Advanced Use and Troubleshooting (90 minutes)

By session three, people have tried it on their own. They’ve hit limitations. They’ve been frustrated by hallucinations. Good. Now we can address real problems:

• How to fact-check AI outputs
• When AI is the wrong tool
• Advanced prompting techniques
• Integration with other workflows

Addressing the Real Concerns

Every team has someone convinced AI will make them redundant. You can’t train around this—you have to address it directly.

Here’s what I tell people: The employees who will struggle aren’t the ones who don’t know AI. They’re the ones who refuse to learn anything new. The skills that matter—critical thinking, relationship building, creative problem-solving—become more valuable when routine tasks are automated.

AI doesn’t replace judgment. It doesn’t replace the ability to read a room. It doesn’t replace twenty years of industry experience. What it does replace is the tedious parts of work that nobody enjoys anyway.

Create Space for Experimentation

One thing that kills AI adoption is making people feel stupid when they get it wrong. I’ve watched senior executives avoid using new tools because they don’t want to look incompetent in front of their teams.

Create explicit permission to experiment and fail. Consider:

• “AI learning time” where mistakes are expected
• Peer learning groups instead of formal training
• Anonymous question channels
• Regular “what I learned” sharing sessions

Measuring Progress the Right Way

Forget about measuring prompts per day or time saved. In the early stages, measure:

• Confidence levels (survey before and after)
• Number of people who’ve found one useful application
• Quality of questions being asked about the tool
• Voluntary sharing of tips between team members

The efficiency metrics will come later. First, you need adoption.

The Manager’s Role

Middle managers can make or break your AI training program. If they’re not using the tools themselves, their teams won’t either.

Before any staff training, run a separate session for managers. Let them voice their own concerns privately. Give them talking points for their teams. Make sure they understand that their role isn’t threatened—it’s evolving.

What Not to Do

A few lessons learned the hard way:

Don’t mandate usage. You’ll get compliance, not adoption.

Don’t track individual usage metrics. Nothing kills experimentation faster than feeling surveilled.

Don’t position AI as the solution to understaffing. Even if it’s true, saying it out loud makes people feel like numbers.

Don’t skip the change management. I’ve seen organisations spend months on technical implementation and zero time on human factors. It never works.

The Long Game

Real AI fluency doesn’t come from a training program. It comes from a culture where learning new tools is normal, where experimentation is safe, and where people genuinely believe their development matters to the organisation. AHRI’s research on workplace learning confirms that cultural factors consistently outweigh training program quality in determining skill development outcomes.

The training sessions I’ve described will get you started. But the real work is creating an environment where people keep learning long after the formal program ends.

That takes time. It takes consistent messaging. It takes leaders who walk the talk. And it takes genuine investment in your people, not just your technology.

The organisations getting this right aren’t the ones with the best AI tools. They’re the ones with the strongest learning cultures. Everything else follows from that.