AI is a bicycle, Not a body

When I was young, I walked to school. Then one day I got a bicycle. The bicycle didn't replace walking. I still walked to the corner shop, still walked when the bike had a flat. But for the school run, it was faster and I arrived less tired.

Then came the bus. Same route, no effort, could read on the way. And in adulthood a scooter for short stretches because it was quicker through the car park.

None of these replaced the others. Each was a medium, a better fit for a specific job.

I keep coming back to that image every time someone says AI is going to replace workers.

The NZ government has announced it's cutting 8,700 public service roles by 2029 and embedding AI "as a basic expectation" across agencies. Those two things are being discussed together, and I think that's worth separating.

The fiscal pressure drove the headcount decision. The AI is the bicycle. Conflating the two makes workers afraid of the tool, and a team afraid of the tool won't use it well.

What AI actually does in a workplace is handle the repetitive, rule-bound, high-volume work so your people can focus on the judgment-intensive work. Chatbots are a clean example. Before modern AI, a reasonable chatbot could handle maybe 40-50% of common customer queries: FAQs, status checks, simple form guidance. Route the rest to a human. The human doesn't disappear. They stop spending half their day answering the same 5 questions.

But, and this is a real but: you have to know who your customer is first.

I've seen organisations roll out chatbots to serve a customer base that barely uses smartphones. The deployment looked impressive. Adoption was close to zero. Because the assumption that most customers are digitally comfortable wasn't true for that particular service.

The question before any AI implementation is simple: who are we serving, and what percentage of them can actually use this?

If 80% of your users engage through digital channels, build for them and keep a human channel for the rest. If that ratio is flipped, the chatbot is the wrong starting point. This is a customer understanding question, and the technology just makes the gap more visible.

The government's plan calls for cloud HR, payroll automation, case management systems. That's the back office. That's the bus route that runs without a driver.

What's harder to address, and what I haven't seen clearly articulated in the plan, is the cultural piece. Technology adoption in organisations doesn't fail because the technology is wrong. It fails because the belief system doesn't shift.

I saw this pattern clearly in a restructure at an education business. The service desk team was managing laptop deployments manually, staging, imaging, dispatching across 48 locations. The capability to automate most of it already existed through the hardware supplier, but the team had been doing it their way for years and didn't fully trust that the supplier's process would hold up.

The turning point was a pilot. One batch, fully outsourced to the supplier, end to end. Cost about $50 per laptop, imaging and logistics included. The batch arrived on time, correctly configured, no issues.

After that, nobody wanted to go back.

That's how adoption actually works. Small wins, specific demonstrations, one thing at a time. The bicycle is faster than walking, but you have to ride it once before you believe it.

So if you're a leader trying to navigate this: pick one process, something specific, bounded, and annoying, and make it your AI pilot. Not a transformation programme. One thing.

Show the team what the bicycle does. Let them ride it.

AI won't replace human creativity and judgment, and it won't replace the person who can read a room, ask the right question, or know when the data is telling you the wrong story. What it will do is take the robotic work off their plate so they have time to do those things.

The medium changes. The destination is still yours to choose.

It's made sense to me.