In the previous posts, I wrote about speaking at Global Azure ANZ 2026 and the strange speed at which universities seem to be reversing themselves on AI.

But the conversation that lingered longest in my head afterwards was about coding itself.

Or maybe not coding exactly.

More about the feeling of coding.

One developer told me they just were not enjoying the new AI coding tools very much. The outputs were fine. Useful even. But something felt missing. The satisfaction had thinned out somehow.

Later that same evening, I spoke with an architect who looked like they had discovered caffeine for the first time. They were ecstatic about vibe coding. Shipping faster. Exploring more ideas. Building things that would previously have sat in notebooks or half-finished repos for months.

Same room. Same tools. Completely different emotional weather.

And I keep thinking about that because it feels like people are talking past each other in these AI discussions. One side talks productivity. The other talks craft. Both think they are discussing software engineering, but they are often defending entirely different experiences of it.

Some people genuinely love the act of construction itself. Sitting with a hard problem for hours. Chasing edge cases through logs and stack traces. Slowly refining an abstraction until it finally clicks into place. There is rhythm in that work. A kind of quiet satisfaction that is difficult to explain to non-technical people.

You disappear into it for a while.

Then emerge later with something elegant.

AI changes that rhythm.

Sometimes beautifully. Sometimes awkwardly.

Code appears too quickly now. Entire structures materialise before your brain has emotionally caught up with them. Useful, yes. Impressive too. But occasionally it feels like skipping straight to the answer section in a maths textbook and missing the strangely rewarding frustration in between.

Not everybody experiences that as loss though.

For some developers, the old friction was never sacred. It was just friction.

Boilerplate. Setup. Glue code. Endless implementation detail sitting between an idea and reality. AI burns through those layers astonishingly fast, which means certain people suddenly feel more creative than they have in years.

You can see it in the way they talk.

Less exhausted. More playful. More experimental.

Like the distance between imagination and execution collapsed almost overnight.

And honestly, maybe it did.

That is what makes vibe coding interesting beneath all the memes and internet snark. It is not really about being lazy or careless. At least not most of the time. It is about interacting with software in a more conversational way. Steering rather than manually carving every piece by hand.

A lot of developers hate that framing.

Others absolutely light up around it.

Which probably tells us something important.

Software engineering was never just one thing to begin with.

For some people it is architecture. For others, craftsmanship. Some love systems thinking. Some love optimisation. Some love building products quickly. Some love understanding every layer deeply enough to feel the machinery underneath their fingertips.

AI amplifies those differences rather than flattening them.

And maybe that is why the discussions get emotionally charged so quickly. Because underneath the tooling debate sits a quieter identity question:

What part of this work actually matters to me?

That question gets uncomfortable fast.

Especially because the AI tools are undeniably useful now. We are well past the stage where dismissing them entirely feels intellectually serious. The leverage is real. The acceleration is real. Small teams can suddenly do things that previously required entire departments and six months of meetings.

But.

Usefulness and meaning are not the same thing.

That tension keeps surfacing everywhere.

A musician using AI mastering tools. A designer generating concepts in seconds. A writer watching language models produce structurally competent prose on demand. Even when the outcomes are good, there can still be a strange emotional wobble underneath it all.

Humans do not only want efficiency.

We also want attachment. Ownership. Struggle. Little moments where effort transforms into identity.

Which makes this whole transition far more philosophical than people expected.

For years the software industry pushed toward higher abstraction. Better frameworks. Faster tooling. Less boilerplate. More leverage. AI simply continued that trajectory with terrifying speed. And now, suddenly, some developers are discovering that removing friction also removes parts of the experience they quietly valued.

Not everybody of course.

Some are having the time of their lives.

Honestly, watching certain architects and senior engineers use modern AI tooling is incredible. They move through systems almost like conductors now. Less typing. More orchestration. Less wrestling syntax. More shaping intent.

That is probably where the industry is heading.

But I also think the uneasy developers are seeing something real too.

There is a danger in becoming disconnected from the reasoning process. A danger in accepting generated outputs without developing the instincts required to challenge them. Shallow understanding scales surprisingly well right until everything catches fire in production.

And production always catches fire eventually.

So perhaps the future is neither full manual craftsmanship nor blind AI delegation.

More likely it belongs to developers who can move fluidly between both modes. People who understand systems deeply enough to reason independently, while also knowing when to lean on AI for leverage.

That balance feels increasingly important.

So does taste.

Taste in architecture. Taste in prompts. Taste in knowing when generated code is elegant versus merely plausible. Taste in recognising hidden complexity before it explodes three months later.

Funny really.

For all the discussion about AI replacing engineers, many of the skills becoming more valuable now seem profoundly human. Judgement. Curiosity. Communication. Restraint.

And maybe that is why the conversations at Global Azure stayed with me long after the event ended.

Not because people were debating whether AI matters anymore.

That debate is over.

The more interesting question now is what kind of relationship people want with the technology shaping their work.

Some want acceleration.

Some want craftsmanship.

Most probably want both.

And all of us are still figuring out where the balance sits.

Written for KiwiGPT.co.nz — Generated, Published and Tinkered with AI by a Kiwi