Fifty-one per cent of New Zealand workers say they learned their AI skills from YouTube and TikTok.
Not 51% of the enthusiasts. Fifty-one per cent of workers, in a survey of 820 employees across businesses from five staff to 1,500.
So the question was never whether your people get AI training. They’re getting it. It’s running right now — on their phones, at home, at ten at night. It has no quality control, no security awareness, no idea what your business does, and no opinion about what “good” looks like in your industry.
The only real question left is whether you get a say in it.
The lunch-and-learn problem
Plenty of New Zealand businesses have tried something. Usually a session: one room, one hour, someone who knows AI walking everyone through it.
Three months later nothing has changed, and the conclusion drawn is that the staff are resistant.
I don’t think resistance is what happened. I think the format was backwards.
A one-hour session gives forty people the same pace and no method. That is precisely the wrong way round.
One method, forty paces
Consistency belongs in the method. How your business decides what to hand to AI. How you check what comes back. What never leaves the building. What gets written down so the next person doesn’t start from scratch. That should be identical for everyone, because forty people improvising forty different standards is the problem you’re trying to solve, not the solution.
Pace belongs to the person. The 24-year-old in marketing and the 58-year-old in dealer support do not need the same runway, and pretending they do is how you lose the second one for good. She doesn’t need the session. She needs one job she does every single week to get noticeably easier, once, with nobody watching her do it.
Most AI training does the exact reverse. Identical pace, no method. Everyone sits the same hour, then goes back to their desk and invents their own rules — which is the state you were already in, except now you’ve spent money on it.
What no method actually looks like
Here’s a number from MYOB’s 2026 Business Monitor that I find genuinely uncomfortable. Small businesses are running legal and contract documents through AI at 32% — roughly double the rate of medium and large firms.
Read that charitably, because I think the charitable read is the true one. Small businesses aren’t reckless. They have less access to a lawyer and more pressure to keep moving. The tool is right there, the need is real, and nobody has ever told them what’s in bounds.
That’s not a discipline problem. That’s an absence of method, filled in by someone doing their best.
And it’s the same shape as the marketing coordinator pasting client details into a free tool, or the ops manager who’s built something brilliant in a chat window that will vanish the day they resign.
The bit that scales
You’ll notice I haven’t said “hire a trainer” or “buy a course.” Courses are events. What people need is closer to a habit, and habits are built on real work, not worked examples.
New Zealand’s enterprise technology leaders landed in roughly the same place this year. Tuanz, after interviewing the technology chiefs of ANZ, Transpower, Kiwibank, Freightways and others, concluded that static IT policies are “no longer sufficient” and that what’s required is rolling, ongoing capability building — teaching people to prompt well, to recognise when the machine is confidently wrong, and to keep a human accountable for the decision.
That’s the enterprise version. The version that works in a forty-person business has to be smaller, cheaper and woven into the work itself, or it simply won’t happen — there’s nobody free to run it.
But the principle survives the translation. Same method, individual pace, on work people actually do, repeated rather than delivered once.
If you do nothing
This is the part I’d want a business owner to sit with. The training happens either way.
Your people will keep learning AI. That’s not in your control and hasn’t been for two years. What’s in your control is whether what they learn reflects your business, your risk tolerance and your customers — or whether it reflects an algorithm optimised for how long someone watches a video.
Right now, for most New Zealand businesses, it’s the video.