Today, a brief exchange on X got me thinking.
In response to a clip showing AI generating a full PowerPoint deck in seconds, Tom Goodwin remarked that the work was never the slides, but everything else around them. Something I’ve always taken for granted, but I’m beginning to realise more and more, that many believe this was the “work".
Because it’s hard not to notice how much modern work has become entangled with the deliverable. The deck, the document, the plan. Somewhere along the way, these artefacts stopped being outputs and became the thing work was organised around. They became how effort was justified, how value was signalled, how progress was made visible
When the evidence of work disappears
When people watch AI generate a PowerPoint in seconds, the reaction is visceral.
Not curiosity.
Not interest.
Something closer to grief.
It’s over.
What’s left for us now?
Why would anyone need this role anymore?
I understand the reaction, because for most of our working lives we’ve been trained to believe that the deliverable is the work. The deck, the document, the strategy, the report. These artefacts became the visible proof that effort had been applied, that value had been created, that a job had been done.
But that belief was always a convenience. A shortcut. A way of making invisible labour legible inside large organisations.
AI hasn’t replaced work.
It has stripped away the evidence.
And that is far more unsettling.
How artefacts became a substitute for value
Somewhere along the way, knowledge work lost its physicality. We stopped making things you could touch. In response, we started producing things you could point to. Slides. Frameworks. Plans. Outputs that looked substantial enough to justify their own existence.
Over time, the artefact stopped being a byproduct of thinking and became the thing that thinking was organised around. Entire careers were built on producing them. Not necessarily because they moved organisations forward, but because they were safe, measurable, and defensible.
They gave the comforting illusion that progress was being made.
AI has made those artefacts cheap.
That’s the moment we’re living through now, the moment when the receipt for labour no longer proves that labour mattered.
Where the real work actually happens
Over the last few months, I’ve been working with a business to fundamentally redesign how it operates commercially. AI was everywhere in that process. It helped with analysis, research, writing, scenario modelling, and the creation of supporting materials.
If you looked only at the artefacts, you might conclude that AI did most of the work.
It didn’t.
The work happened in the conversations that followed. In the moments where people had to let go of familiar structures they’d built their identities around. In the slow process of helping teams see the system they were operating inside, and the one they were about to enter. In the resistance, the anxiety, the politics, the loss of certainty.
I spent far more time persuading, mentoring, translating, defending, and sequencing than I did producing anything tangible.
The people I was working with had access to the same AI tools. They could generate the same outputs. They still couldn’t have done the work.
Not because they weren’t capable.
But because AI doesn’t carry responsibility.
The jobs that are already disappearing
This is where the story becomes uncomfortable, because it’s also true that AI is destroying jobs. That’s not a future scenario, it’s already happening.
Roles built around execution, assembly, and first-draft production are being compressed or removed entirely. Customer support teams shrink as deflection rates climb. Content roles disappear as one person with AI replaces what used to be a small team. Analyst roles vanish even as the need for insight remains.
Organisations talk about productivity, but what they often mean is headcount reduction. The gains accrue upwards, not outwards.
And quietly, something else is happening too.
The disappearance of the on-ramps
The entry points are going.
The junior roles that once existed partly because work was inefficient, the roles where people learned by doing, by producing, by making imperfect things, are being erased. AI doesn’t just remove jobs. It removes apprenticeships in disguise.
We are accelerating toward a future where fewer people know how the system actually works, even as the system becomes more complex.
Why this isn’t a contradiction
None of this contradicts the idea that the deliverable was never the work.
It proves it.
The roles under pressure are the ones whose value was defined by outputs rather than outcomes, by artefacts rather than accountability. Those roles were always vulnerable. AI has simply made that vulnerability visible.
Meanwhile, the work that remains stubbornly resistant to automation looks very different. It lives in judgement. In sense-making. In the ability to hold ambiguity without rushing to false certainty. In knowing when not to act. In carrying the consequences of a decision once it leaves the slide deck and enters the organisation.
AI can generate options.
It cannot own the outcome.
The limits of automation
Some argue this is only a temporary gap, that AI will soon simulate leadership, persuasion, even systems thinking better than humans ever could.
Perhaps.
But organisations don’t adopt capability at the speed it becomes possible. They adopt it at the speed they can tolerate risk, blame, and loss of control. There is always a lag between technical potential and human acceptance.
That lag is where much of today’s value is being created.
Others argue that this framing simply protects senior people and entrenched power. Sometimes that’s true. But the answer isn’t to pretend that judgement doesn’t matter. It’s to redesign work so that judgement is visible, shared, and developed, not hidden behind artefacts and titles.
Because organisations that continue to hire for deliverables, measure output, and reward speed over understanding won’t become more efficient.
They’ll become brittle.
What remains
The future of work is not a clean story about humans versus AI.
It’s messier than that.
AI will continue to eliminate jobs. It will also make the remaining human work harder to fake. There will be fewer slides, fewer disguises, fewer places to hide.
What remains will feel heavier. More exposed. More human.
And that may be the most honest version of work we’ve had in a long time.
