AI Training for Employees: Why So Few Feel Trained, and What Actually Works

AI training works when people practise on their own real tasks, get enough hours to build the habit, see leaders using AI, and have a shared setup to come back to after the session. Most training fails because it is a one-off demo of generic prompting. The gap is less about knowledge than about practice and support.

How big is the AI training gap?

Large. Most employees now use AI in some form, but few have been taught to use it well. Surveys consistently find that only around a third feel properly trained, and fewer than half of AI users have received any training from their employer. Frontline staff are least supported.

According to BCG's 2025 AI at Work survey of more than 10,600 people, only one-third of employees say they have been properly trained. Frontline employees' regular use had stalled at 51%, and only about a quarter of frontline staff said they get enough leadership support.

The Microsoft and LinkedIn 2024 Work Trend Index found a similar gap from another angle. Three-quarters of knowledge workers use AI at work, but only 39% of AI users have received AI training from their employer. Most people taught themselves.

Why does most AI training not stick?

Because it is usually a one-off session on generic prompting, disconnected from people's actual work. Staff watch a demo and try a few examples. Then they return to their jobs with no shared setup, no follow-up and no clear rules. Within weeks, a few enthusiasts keep going and everyone else drifts back to old habits.

Four common failures:

Generic content. A session on "how to write a good prompt" does not show a finance officer how to check an invoice batch with AI.

Too little time. A single hour builds awareness, not habit.

No follow-through. Questions come up in week two, when there is nobody to ask.

Nothing to build on. Each person goes back to a blank assistant that knows nothing about the business, so they rebuild context alone.

What does good AI training look like?

Good AI training is hands-on and uses people's own tasks. Sessions are grouped by role, delivered in small groups and followed by ongoing support, such as a regular clinic. Good training covers the rules on data and approval as well as technique. It ends with each person knowing what they will use AI for on Monday.

A useful structure:

1. Leaders first. Leadership agrees where AI fits and what it may touch, then models using it.
2. A company kickoff. What is changing, what is not, the rules, and where to get help.
3. Team workshops. Each team works through its own real tasks, hands on.
4. A monthly clinic. Drop-in help for questions that come up in practice.

The rules belong in training too. Australia's National AI Centre lists maintaining human control among its six essential practices for AI adoption, and staff need to know where that control sits in their own work.

Leadership matters more than most training plans assume. According to BCG, the share of employees who feel positive about generative AI rises from 15% to 55% with strong leadership support.

How much AI training do people need?

More than an hour. BCG found that regular use is sharply higher among employees who receive at least five hours of training, especially with access to in-person training and coaching. Treat five hours as a practical minimum, spread across a kickoff, a hands-on workshop and follow-up support.

Spreading the hours matters as much as the total. A single long session overwhelms people; a workshop followed by short clinics lets them try things on real work and come back with real questions. The guide to running an AI workshop covers how to structure the hands-on part.

How do you keep AI skills growing after training?

Give people a shared setup that does some of the work for them. That means business knowledge their AI reads, written workflows for common jobs, and a place to share what works. Name a champion in each team with time to help colleagues. Training then becomes the start of a habit rather than a one-off event.

The shared setup is underrated. If every person's AI already knows the product facts, customer language and brand voice, people get good results without being expert prompt writers. The skill gap narrows because the system carries part of the load. That shared knowledge is what a company brain holds.

Champions help too. The guide to AI champions covers how to choose them and protect their time.

Adoption is one of Skein's three ways in, alongside strategy and implementation, and the AI facilitation page shows how it works on a team's real work. That suits a business where licences were bought for everyone and are used by a few. A team that is already using AI well may only need the clinic.

Frequently asked questions

How do you train employees to use AI?

Start with leaders agreeing the rules and modelling use, then run a company kickoff and hands-on workshops where each team practises on its own real tasks. Follow up with a monthly clinic and named champions. Give everyone shared business knowledge their AI reads, so good results do not depend on prompting skill.

How much AI training do employees need?

BCG's 2025 AI at Work survey found regular use is much higher among employees who receive at least five hours of training, particularly with in-person coaching. Treat five hours as a practical minimum, spread across a kickoff, a hands-on workshop and ongoing follow-up support.

Why don't employees use the AI tools we bought?

Usually because they were not trained on their own work, have no follow-up support, lack clear rules, or find the tool knows nothing about the business. BCG found only one-third of employees feel properly trained. Hands-on training, leadership support and shared business knowledge address all three.

What should AI training for staff cover?

It should cover practical use on each team's real tasks, the company's rules on what data may go into which tools, where a person must approve output, how to check AI work for errors, and where to get help. Generic prompting tips are far less useful than role-specific practice.

Who should deliver AI training in a small business?

A mix works best: an experienced facilitator for the kickoff and workshops, leaders who model use, and internal champions in each team who help colleagues day to day. The facilitator builds the habit; champions and a monthly clinic keep it going after the sessions end.

Related reading