Sure - eventually, some jobs will disappear. Some roles will shrink. Some will evolve into something unrecognisable.
But I think the immediate impact of AI is being misunderstood.
The first divide won't be between AI and non-AI workers. It'll be between people who know how to apply the new tools to solve old problems… and people who don't.
The First Winners Won't Be the Best Prompt Engineers
Right now, we're still in the early phase of AI adoption. The winners aren't necessarily the people who can craft the cleverest prompts, or generate the fanciest images.
They're the people who can take a messy, frustrating, real-world problem and say:
- What's the actual goal here?
- What does "good" output look like?
- What data do we have? What data do we need?
- What are the edge cases and failure modes?
- How do we validate the result?
That skillset isn't new. It's the same thinking that makes good engineers, good analysts, good consultants, and good operators.
AI just gives them a bigger lever.
The Ladder of Automation
At the bottom of the ladder are the obvious wins — fixing the day-to-day annoyances everyone complains about but nobody prioritises:
→ Repetitive reporting
→ Copying data between systems
→ Manual reconciliation
→ Generating status updates
→ Cleaning up spreadsheets
→ Producing docs from templates
But higher up the ladder, you stop improving the workflow — and start eliminating it.
And then you're not automating a task anymore — you're automating a role.
That's where the job impact becomes real. Not because AI is "intelligent", but because it makes it economically viable to build systems that previously weren't worth the effort.
AI Is Not a System
One of the biggest mistakes I'm seeing is people trying to use AI as if it is the system.
The mindset goes something like: "Let's just give the spreadsheet to AI and see what it recommends." Or: "Let's just ask the AI what we should do."
That approach sounds exciting, but it collapses quickly in real-world environments.
Because organisations don't want "an answer". They want consistency. They want reliability. They want auditability. They want the same inputs to produce the same outputs — or at least predictable ones.
In other words: they want a system.
AI can be a component inside a system, but AI on its own isn't one. It's not deterministic. It doesn't have guarantees. It doesn't understand your business context unless you explicitly provide it. And it won't magically infer missing information just because it sounds confident.
If a human couldn't generate a meaningful recommendation from a spreadsheet with no context… neither can AI. You'll just get a well-written guess.
The Real Shift: Systems Just Got Cheaper
This is the part I think people are underestimating.
AI doesn't just automate tasks. It drastically reduces the cost of building automation.
The "first draft problem" has basically disappeared. The friction of going from idea to prototype, prototype to workflow, workflow to product, product to automation — is collapsing.
And when the cost of building systems drops, the ambition level rises. Projects that would have been rejected as "too expensive" or "not worth the engineering effort" suddenly become viable. That changes the shape of organisations. Not overnight, but quickly.
The Future Isn't "Ask the AI to Run the Business"
The most effective future isn't one where people just ask AI what to do and hope for the best. The future is systems that encode repeatable processes, automation that removes low-value manual work, AI used selectively where traditional algorithms struggle, and humans still responsible for defining goals, constraints, and outcomes.
AI is most powerful when it sits at the boundary between structured and unstructured problems — where it's too difficult to write a rigid algorithm, where you need interpretation, judgement, summarisation, classification, or reasoning under ambiguity. That's where it shines. But the rest of the workflow still needs to be engineered.
So What Happens to Jobs?
There will definitely be a divide. But at least initially, the divide won't be between jobs that AI can do and jobs that it can't. It will be between people who can use AI to build leverage… and people who keep trying to use it like a magic oracle.
In the short term, the people who thrive will be the ones who can turn AI into systems.
And then it becomes a race: how high up the ladder do you want to climb? Do you stop at eliminating daily frustrations? Or do you aim for automating the entire process — and removing the need for the role entirely?
AI doesn't answer that question. People do.
Want to talk about where AI fits in your organisation?
We help businesses build systems — not just experiments.