For years, the dominant narrative around AI has been simple: machines are coming for jobs. We’ve heard this framing so often that it’s become background noise, but it rests on a flawed assumption.
The assumption is that jobs are fixed bundles of tasks, and if you automate enough of those tasks, the job itself disappears.
That’s not how work actually functions.
As Jensen Huang recently articulated, a job isn’t defined by the tasks. A job is defined by its purpose. The tasks are just implementation details.
Once you grasp this distinction, everything about the AI-and-work conversation begins to look very different.
Jobs Were Never About Tasks in the First Place
Ask someone what a lawyer does, and they’ll tell you:
- they write contracts
- they research case law
- they prepare arguments
- they draft legal documents
Ask a marketer, and you’ll hear about writing copy, running ads, analysing metrics, testing campaigns.
Ask an engineer, and they’ll mention code, debugging, documentation, building features.
This is how we’ve always talked about work. We inventory the activities, we make lists, and for a long time, it’s been very useful in describing what we do.
However, If you zoom out slightly, you realise that none of these things are actually the job. They’re all intermediate steps.
A law firm doesn’t hire lawyers to produce documents. If that were the job, why would they exist now that document templates exist?
The reason law firms exist is because clients need someone to navigate a legal system and protect their interests.
The documents are just evidence of that work. They’re the artifact, not the purpose.
The same applies to marketing.
A company doesn’t hire a marketer to write ads. The ads are the output.
What they’re really paying for is someone who can understand a market and influence people toward a desired outcome.
And with engineering, no one cares about the code itself. They care about the business problem the code solves.
Once you separate activity from purpose, the AI-and-jobs conversation becomes clearer.
AI Does Not Replace Jobs — It Removes Task Friction
AI is objectively good at certain things:
- drafting
- summarising
- retrieving information
- generating variations
- structuring data
- rewriting text
All of these live in the execution layer. They are the how of getting work done.
But there are parts of work AI struggles with:
- defining intent
- deciding what the goal should be
- taking responsibility for outcomes
- making judgement calls under ambiguity
- being accountable when things go wrong
That’s where human value concentrates.
So what actually happens is this:
AI compresses the execution layer.
It shortens the distance between deciding what needs to happen and producing the output.
This is where people misread the situation.
They see AI doing the work and assume the job is obsolete.
But what actually happens is that the job changes.
The friction that used to consume most of a professional’s time disappears, forcing them to move upstream—closer to the purpose of the role.
Work Is Defined by Purpose
Take the legal profession.
For decades, being a good lawyer meant:
- drafting agreements
- researching precedent
- writing legal memos
These are valuable skills but they are not the job.
The job is:
achieving the best outcome for a client within a legal system.
Now introduce AI.
It can:
- draft agreements
- retrieve relevant precedent
- summarise case law
Suddenly, a lawyer’s day looks different.
Instead of spending hours producing documents, they spend more time:
- reviewing outputs
- advising clients
- negotiating outcomes
- making strategic decisions
What changed?
The lawyer didn’t lose their job.
The job got compressed.
Execution was reduced.
Purpose remained.
The Hidden Effect: Judgement Becomes More Valuable
When execution friction disappears, something else becomes the bottleneck:
judgement.
Not the ability to do something—but the ability to decide what should be done.
Judgement involves:
- understanding context
- identifying constraints
- choosing desirable outcomes
- making trade-offs
This cannot be easily systematised.
It becomes more valuable as execution becomes cheaper.
When anyone can produce output, the scarcity becomes:
knowing what to produce.
This is why AI does not flatten skill hierarchies.
It reshapes them.
The Shift From Doers to Directors
For most of the 20th century, value came from execution.
The best writer, engineer, or analyst had an advantage because execution was scarce.
AI changes that.
Execution becomes cheap.
So value shifts upward:
- from doing → directing
- from producing → deciding
- from output → outcome design
This introduces new high-value skills:
Problem Framing
The ability to clearly define what needs solving.
System Thinking
Understanding how different parts of a workflow interact.
Decision Design
Structuring choices to produce consistent outcomes.
Context Engineering
Providing the right inputs so systems produce useful outputs.
These are not new skills,they were just hidden before.
Why High-Skill Workers Become More Valuable
A common fear is that AI levels the playing field.
It doesn’t.
AI removes baseline effort, but it amplifies thinking.
A mediocre thinker with AI becomes slightly more productive.
A strong thinker with AI becomes exponentially more effective.
AI amplifies direction—not intelligence.
If you know what you want, AI accelerates you.
If you don’t, it just produces confusion faster.
The Real Bottleneck Has Moved
Before AI:
- writing took time
- research took time
- analysis took time
Execution was the bottleneck.
Now:
- drafts take minutes
- research is instant
- outputs are immediate
The bottleneck moves upstream.
It becomes:
- clarity of intent
- quality of problem framing
- strength of judgement
AI is not inconsistent.
It is extremely consistent with the input it receives.
Good input → good output
Confused input → confused output
Why “AI Will Replace Jobs” Is the Wrong Question
The real question is:
What part of a job is actually the job?
If you remove all tasks, what remains is:
- purpose
- responsibility
- judgement
- direction
These are difficult to automate.
Because they are context-dependent.
And context is where humans still dominate.
What This Means for Professionals
If you work in law, marketing, engineering, writing, or analysis, your job is not disappearing.
But it is changing.
You will spend less time:
- producing outputs
- doing repetitive tasks
And more time:
- reviewing work
- making decisions
- defining goals
- structuring systems
This is a shift from producer to director.
And that’s where the real value has always been.
The New Competitive Advantage
In an AI-driven world, the most valuable skills are cognitive:
- clarity of thinking
- precision in defining goals
- structuring ambiguity
- guiding systems toward outcomes
These skills were always important.
But now they are visible.
And that visibility creates a new hierarchy.
Conclusion: AI Removes Tasks, Not Purpose
Jensen Huang’s insight is simple:
A job is defined by its purpose, not its tasks.
AI removes the how.
It does not remove the why.
The lawyer still delivers legal outcomes.
The marketer still drives results.
But they no longer spend most of their time on execution.
What remains is the part of work that matters most:
- judgement
- direction
- decision-making
These are the things AI does not replace.
It makes them more valuable.
Top comments (0)