Most AI training for managers teaches tools, and tools change every six months. The training worth buying teaches judgement: when to use AI, when not to, how to check its output, and what your team is allowed to put into it. If you are briefing a provider in Hong Kong, ask what people will be able to decide differently afterwards, not which products they will have been shown.
Over the past year, “we need to do something on AI” has become one of the most common requests landing on HR desks in Hong Kong. It usually arrives from the top, it usually arrives without a brief, and it usually arrives with a deadline.
The problem is that the market has filled up faster than it has matured. A lot of what is sold as AI training is a demo. Someone walks a room through a chat interface for ninety minutes, everyone leaves impressed, and nothing changes in how the business works. Six months later the tool has been replaced and the training has aged out with it.
Here is how to tell the difference before you buy.
Why does most AI training for managers fail to stick?
The first reason is that it teaches the product rather than the decision. A session built around one vendor’s interface is obsolete the moment that interface changes, which it will. What does not go out of date is knowing which tasks are worth handing to a machine, which are not, and how to tell when the output is wrong.
The second reason is that it ignores the thing managers are actually anxious about. In my experience the questions that come up in the room are rarely “how do I write a better prompt”. They are closer to “am I allowed to put this document into it”, “how do I know if my team is using this already”, and “what do I do if it produces something confidently wrong in front of a client”. Training that does not address those leaves people more cautious than when they walked in.
The third is audience. AI training sold to a room of thirty people spanning finance, marketing and operations is a talk, not training. The work that changes behaviour is narrower.
What should AI training for managers cover?
Four things, in roughly this order.
Judgement about where it helps. Which parts of your managers’ week are genuinely suited to AI assistance, and which are not. This is specific to the business, which is why a provider should be asking you questions before they propose anything.
Verification. How to check output that is written to sound correct. This is the single most transferable skill in the whole subject and the one most often skipped.
Governance in plain language. What your people may and may not put into a tool. Most organisations in Hong Kong have either no policy or a policy nobody has read. A provider who asks to see yours before designing the session is doing it properly. If you do not have one, a good provider will say so rather than working around it.
Practice on real work. Generic exercises produce generic confidence. The session should use the team’s own tasks, which means the provider needs to understand the work before the day.
What should you ask a provider before you book?
A short list that separates serious providers quickly.
- What will managers be able to decide differently the week after this session?
- How do you keep the material current when the tools change?
- Whose policy and risk position are you designing against, ours or a generic one?
- What do you need from us beforehand, and how long will that take?
- Who is facilitating, and what have they actually run in an operating business?
That last one matters more in this subject than in most. AI training has attracted a lot of people who are fluent in the technology and have never had to manage anyone through a change in how they work.
Does AI training need to be Hong Kong specific?
Partly, and it is worth being precise about which parts.
The judgement and verification content travels. It is the same whether the room is in Hong Kong, Singapore or anywhere else.
What does not travel is the governance layer. Data handling expectations, what is acceptable to put into a third-party tool, and the regulatory position of your sector are local questions, and a provider who answers them from a generic template is a provider to be careful with. If your organisation operates across Hong Kong and the mainland, or across Hong Kong and Singapore, say so in the brief. It changes the design.
Language is the other local factor. If some of your managers would rather work through the material in Cantonese or Mandarin, that needs to be in the brief, not raised on the day. Plenty of providers can do it; few will offer if they are not asked.
In-person, virtual, or a programme?
A single session, in person, works well when the goal is a shared starting point across a team that works together.
Virtual works better than people expect for this subject specifically, because everyone is already at the machine they would be using.
A programme spread over several weeks is the right shape when you want behaviour to change rather than awareness to rise. It gives people a gap in which to try something, fail at it, and come back with a real question. It costs more and it is usually the one that produces something you can point at afterwards.
Most organisations buying for the first time should probably start with one well-designed session and decide afterwards, rather than committing to a programme built on assumptions about what the team already knows.
How should you brief it?
The same rules as any training brief apply, and we have written about those in our guide to writing a training brief that gets better proposals. The two additions specific to AI are:
State what your people are currently doing. If half the team is already using AI tools informally and nobody has said so out loud, that is the most important sentence in your brief. It changes the session from an introduction into a correction.
State your constraint. If there is a tool your organisation has standardised on, or one it has banned, say which and why. A provider designing without that is designing something you cannot use.
Frequently asked questions
What is the difference between AI training and an AI workshop?
In practice the words are used interchangeably by providers. What matters is the design, not the label. Ask how long it runs, whether people work on their own tasks, and what happens afterwards.
How much does AI training for managers cost in Hong Kong?
It varies widely with format, length and how much bespoke design is involved, and any provider quoting a figure before understanding your situation is quoting a template. Give a budget range in your brief and let providers design to it.
Should we train managers or everyone?
Managers first, in most cases. They set what is acceptable for their teams, and they are the ones who will be asked the awkward questions. Rolling out to everyone before managers have a position tends to create inconsistency.
How often does this need refreshing?
If the training was built around tools, roughly every time the tools change. If it was built around judgement and verification, considerably less often. That difference is a useful test of what you were sold.
Getting the brief right is most of the job. Our guide to writing a training brief that gets better proposals covers the six elements that turn a vague request into proposals worth comparing, and it applies to AI training as much as anything else.
Stuart Harris is co-founder of Growth Academy Asia. He has spent over 20 years working with HR teams and training providers across Hong Kong and Asia.
