What if the real disruption in AI isn’t that machines are becoming more capable but that we’re quietly redefining work in ways that make humans less so?
It’s an uncomfortable question. But it sits at the center of nearly every conversation about productivity, automation and the future of work.
Across industries, work is increasingly framed in terms machines can understand: outputs, efficiency gains, throughput. Are these necessary metrics? Sure. But they’re not neutral. They shape how we define value – and, by extension, how we define what people are for.
And that’s where the risk emerges. Not from AI itself, but from a subtle shift in perspective: When we begin to treat the output as the work, rather than the evidence of it.
When output becomes the point
In the AI Myths and Mythos episode of the Pondering AI podcast, professor and AI researcher Eryk Salvaggio points to a growing disconnect in how we think about productivity. Systems designed to generate content or complete tasks invite a seductive assumption – that if the output is produced, the work is done.
But Salvaggio points out that framing misses something essential. Giving an example of using AI to draft a government accountability report, the concern isn’t accuracy or speed – it’s purpose. The thinking, reflection and responsibility that should shape the report are bypassed.
“It becomes this kind of weird exercise in the illusion of accountability,” he says. “It’s not real, right? It’s not the real work.”
The public sector is hardly the only discipline where the value has never been the artifact alone. A legal argument isn’t valuable because it exists – it’s valuable because of the reasoning behind the way it’s constructed. A business strategy deck isn’t useful because it’s presentation-ready – it’s useful because it reflects actual human insight and judgment.
When we optimize only for outputs, we risk flattening these contributions into something far more mechanical and far less meaningful.
The work behind the work
In the Pondering AI episode The Nature of Learning, author and educator Helen Beetham brings this argument to a field where this type of tension tends to surface well before workforces are worried about it – academia.
She notes that when students are asked to produce essays, presentations or code, the goal is transformation – to challenge thinking, develop new capabilities and build understanding.
“We’re not doing that so there can be more content in the world,” she says.
That insight translates directly to the workplace. The deliverable is not the work itself; it’s a byproduct of the work. The real value lies in the thinking, questioning and learning that produce it.
When systems allow us to skip directly to the output, we gain speed. But we risk losing something harder to measure and easier to overlook: development. And over time, that trade-off compounds.
Solving the wrong problems
In the Righting AI episode of Pondering AI, acclaimed international human rights lawyer Susie Alegre highlights a different kind of misunderstanding – not just about work, but about what humans value more broadly.
Reflecting on reactions to generative AI, she notes that much of the anxiety among creatives wasn’t simply about replacement. It was that the people embracing this technological revolution did not really understand what human creativity was about and why people create.
“Creativity is the work,” she explains. “It’s not about the product.”
We know AI is good at spotting patterns, but I’m sure you, the (hopefully) human reader, are picking up on the one we’re seeing here whether it’s in business, government, education or art.
Technology can – and should – help reduce burden. But not at the cost of replacing the very parts of experience people value most: connection, creativity and care. (All while, ironically, many of the tedious, repetitive or emotionally draining aspects of work remain unchanged.)
What matters is which parts of work we are making more efficient – and what we preserve.
The quiet shift in how we value work
Dr. Christina Colclough, in the Keeping Work Human mini-episode of Pondering AI, takes this further, connecting these shifts to broader changes in how we think about labor itself. She observes that we’ve “sleepwalked” into a model where efficiency and convenience dominate decision-making, often without fully interrogating their implications.
At the same time, she points out that much of the policy and governance attention around technology focuses on markets – how systems perform, scale, and compete – rather than on workers and the evolving nature of work.
“We are regulating the market far more than we’re regulating the labor market in relation to digital technologies,” she says. “Yet, there can be no economy without the labor market. So there’s a lot of weird stuff going on here.”
The result is a gap. As tools become more powerful, the frameworks that define and protect meaningful work struggle to keep pace. And without that balance, there’s a risk that the idea of what constitutes work and who performs it becomes narrower.
Are we designing for humans – or around them?
None of this is to say that automation is inherently problematic. The ability to reduce friction, accelerate workflows and augment human capability is genuinely transformative.
But transformation doesn’t always mean “better.”
If we define work narrowly – as the production of outputs – then automation will continue to push us toward efficiency at all costs. But if we recognize that work also includes thinking, learning, creating and connecting, then the goal shifts.
Before it’s too late, we need to ask ourselves: Are we designing systems that support human work – or redefining work to fit the systems we’ve built?
How we choose to answer will determine whether we have merely a more productive future or a more human one, too.
