Teaching AI to nervous staff: four mistakes and what to do instead
Start with their own work. Ask nervous staff to brief the AI tool the way they would brief a new starter, answer job worries honestly, stop the manager answering everything, and finish with a prompt they will use the next day.
This is for anyone teaching AI to nervous staff. That might be branch staff, an operations team, or a mixed room where a few people use AI tools every day and the rest have never opened one.
Nervous staff usually go quiet for reasons the session itself creates. Below are four mistakes we see when AI prompting training is run for people who are uneasy about it, and what to say or do in the room instead.
They come from our AI prompting workshops, where a mixed team works on a real customer problem from their own work.
Mistake 1: deciding your staff are not tech people, so prompting is not for them
On a normal working day this shows up early. A manager says the branch team is not very technical, so the AI training goes to head office first. Or someone in the session says "I'm not a tech person" and spends the next hour watching the person next to them.
Writing a prompt is mostly explaining a task clearly. Branch staff and operations people do that all the time. If someone can brief a new starter, they can write a prompt.
So when someone says they are not a tech person, ask them to explain a task to the AI tool the way they would explain it to a new starter. Take something they do every week, like replying to a customer who wants to know why their account opening is taking longer than expected. They will naturally say who the customer is, what the customer already asked, what tone to use and what not to promise.
Then read the answer with them. Where it is wrong, ask what they would tell the new starter to change. That correction is their second prompt.
Mistake 2: reassuring people when they ask if AI will take their job
Someone asks it at the break, or right at the start. "Is this going to replace us?" The instinct is to say no, don't worry. The room hears a promise that nobody in the room is in a position to keep, and the worry stays.
An honest, specific answer calms people down more than a promise does. Say what you actually know about which tasks the AI will help with. If you do not know, say that too. People can tell the difference between a straight answer and a comforting one.
Then have them try it on a piece of their own work, so they can see what it does. A customer email they wrote last week is a good place to start. They will usually see the AI get the structure and tone roughly right, and miss the details only they know, like the branch's cut-off time or what the customer said on the phone yesterday.
Seeing that for themselves settles more nerves than anything the facilitator can say.
Mistake 3: putting the manager in the room and letting them answer everything
Having the department head sit in feels like a signal that the training matters. Then every question gets answered by the manager, quickly and confidently. The team watches. The nervous ones decide it is safer to say nothing than to try a prompt that might go wrong in front of their boss.
When the manager answers every question, the rest of the team stops trying.
Keep the manager in the room, but give them a task of their own. Ask them to write the first prompt on screen, mistakes included. When the answer comes back vague or wrong, everyone sees that a bad first attempt is normal, even for the most senior person at the table. That does more for the quiet people than any speech about it being a safe space.
Then put the next question to someone else at the table, by name. Agree this with the manager before the session starts, so it does not feel like a correction in front of their team.
Mistake 4: assuming everyone will use AI at work once they have done the training
The training gets done. The attendance sheet is full, the feedback scores are good and everyone gets a certificate. On Monday, nobody opens the AI tool, because nothing in the session connected to the work waiting on their desk.
Training that people are made to attend rarely changes how they work on Monday.
Build the session around a real customer problem the team already has. For a branch team that might be the same questions about loan status that come in every day. Spend the session writing and fixing prompts for that problem, and finish with one prompt each person will use the next day.
The test is whether they use it on their next piece of real work, and whether it helps. That is only useful if you know where you started, so write down how long the task takes before the session. Most teams have no before number, and then the question of whether the training worked gets answered with a shrug.
What to do this week
- Pick one real task your team does every week, such as replying to a common customer question, and use it as the exercise.
- Write down how long that task takes today, so you have a before number to compare against later.
- Ask the most nervous person in the team to brief the AI tool the way they would brief a new starter, then read the answer together.
- Agree with the manager beforehand that they will write the first prompt on screen and hand the next question to someone else.
- Write down what you actually know about which tasks AI will help your team with, so you have an honest answer ready when someone asks about their job.
Questions people ask
How do you get staff who say they are not tech people to write a prompt?
Ask them to explain a task to the AI tool the way they would explain it to a new starter, then read the answer with them. If they can brief a new starter, they can write a prompt.
What should you say when staff ask if AI will take their job?
Say what you actually know about which tasks the AI will help with, and say so if you do not know. Then have them try it on a piece of their own work so they can see what it does and what it misses.
Should the manager attend AI prompting training with their team?
Yes, but give them a job. Have the manager write the first prompt on screen, mistakes included, then put the next question to someone else at the table.
How do you know if AI prompting training worked?
Check whether people used the AI tool on their next piece of real work and whether it helped. Write down how long that task takes before the session so you have something to compare against.
On-Off Group trains teams, tests products with real customers, finds where a transformation has stalled and builds what gets it moving, for banks, insurers and enterprises in the Philippines, since 2015. Who we are.
