Could your messy workplace save you?
ChatGPT can ace elite law school exams, write better essays than grad students, and leave professors genuinely unsure if they’re grading a machine or a human. But more than two years into the generative AI boom, the predicted white-collar job apocalypse? Nowhere to be seen.
No tidal wave. No Terminator. Just… mild disruption.
So what’s going on?
This paradox has been debated for a while, and in my own work with clients, it’s been clear for some time. Now, the data is finally catching up.
When you line up employment data against the jobs supposedly at high risk of AI automation, something strange happens: most of those “endangered roles” haven’t actually lost jobs.
🧾 Accountants? Still there.
💼 Lawyers? Still talking and typing.
📇 Data entry workers? Surprisingly untouched.
But… two professions have taken a hit: writers and computer programmers.
Why these two?
It’s not about how “hard” a job is. AI doesn’t care if you have a PhD or barely string a sentence together. What it struggles with is something deeply human: chaos.
AI doesn’t fail because tasks are intellectually complex — it fails when the workflow is messy. If a job requires juggling fragmented tasks, shifting priorities, and reacting to ambiguity — basically, being a manager on a Monday — AI still flounders.
Let’s break it down:
- A lawyer manages a flood of tasks, calendars, calls, and humans.
- An office junior is running from the copier, to a coffee run, to a crisis.
These roles are fluid, dynamic, reactive. AI? It can handle all the complexity in the world — as long as it’s on a single playing field. It hates changing fields. It hates fluidity.
But writers and coders? That’s a different story.They’re often solo workers with well-defined inputs and outputs. Tasks like “write me a media release for our downturn in sales due to Trump tariffs and expected future layoffs” or “Generate Python code that does X” are exactly what LLMs were built for.
Plus, there’s the gig economy twist:
Freelancers are easy to replace, because they are task oriented. Just plug and play.
The very traits that once made tech workers and content creators “future-proof” — autonomy, asynchronous workflows, digital deliverables — may have actually made them easier to automate.
Here’s the irony:
The Silicon Valley gospel of self-optimisation and hyper-efficiency didn’t future-proof jobs — it made them fragile.
Meanwhile, the receptionist with seven browser tabs open, three phones ringing, and a half-eaten sandwich? – They’re still in the building.
So what does this mean for you?
If you want to stay ahead of AI, lean into the parts of work it hates:
- Ambiguity
- Interaction frequency
- Human nuance
- Mess
Above all, remember: AI struggles most with changing playing fields.
Moving from a laptop to an app, to a warehouse, to a boardroom… the more places and people you interact with in a day (or a week), the safer your work is.
Yes, you’ll use AI — all day, every day — but it won’t usurp you.
What AI really struggles with is moving between tasks and shifting goalposts.
It’s great at creating pieces, but not at stitching pieces together from physically different worlds.
It turns out, the future of work probably isn’t about optimisation — it’s about managing the chaos of humanity: the twists, the turns, the beautiful mess.
The future of work may just belong to the brilliantly, usefully disorganised.
Keep Thinking,
Steve.