Who Gets to Decide What AI Does to the Economy
From Aristotle to Marx, people have imagined machines ending inequality. Now that automation is actually arriving, the question is whether society shapes it or just absorbs it.
Key points
- Aristotle described a fantasy in which self-operating machines would make human labour redundant.
- Karl Marx, in 1858, predicted that shared knowledge spread by automation could undermine the foundations of capitalism.
- Today's AI systems are making that centuries-old thought experiment feel suddenly practical.
- How the economic shift plays out depends heavily on political choices, not just technology.
The machines are not coming. In many workplaces, they are already here. What nobody has settled yet is who benefits.
That question is older than most people realise. Aristotle, in his Politics, imagined tools that could act on their own, concluding that if shuttles could weave by themselves, master-craftsmen would have no need of assistants. He framed it as a fantasy. Most economists today treat it as a planning problem.
Karl Marx returned to the same idea in 1858. He wrote about what he called "general intellect", a state where knowledge, once privatised in individual skills and intellectual property, becomes socialised across all of society. When that happens, he argued, the economic logic holding capitalism together starts to crack.
Why does any of this matter now?
Because large language models, the technology behind chatbots like ChatGPT and Claude, are forcing the question from philosophy into budgets and legislation.
The Guardian reported this week on how that intellectual history shapes the current debate. What strikes me, covering this beat, is how rarely the policy conversation catches up with the scale of what is being proposed. Governments are still arguing about disclosure labels on AI-generated content while the structural economic shifts are already moving.
Automation has always promised liberation and delivered it unevenly. Steam looms didn't free weavers; they mostly displaced them. The gains pooled at the top until political pressure, sometimes decades later, redistributed them through labour law and public services. Our September story on UK graduate job losses showed coding and finance roles thinning out already, which suggests this wave isn't waiting for the policy debate to catch up.
There's no technical reason AI must follow that same pattern. Nothing guarantees it won't.
What does this mean for ordinary workers?
It depends almost entirely on decisions being made right now in legislatures and boardrooms, not in research labs.
If productivity gains from AI flow mainly to shareholders, people whose tasks get automated will face lower wages or redundancy without obvious replacements. Tax those gains, share them through shorter working weeks or retraining, and the picture looks quite different.
Neither outcome is locked in. That's the actual point: the technology doesn't decide this. People do, through the choices their governments and employers make in the next few years. Watch the policy fights over AI taxation and labour classification. Those are the real test of whether this wave breaks differently from the last ones.
Common questions
Is AI actually going to take most jobs?
Researchers disagree. Most studies suggest AI will change the tasks inside many jobs rather than eliminate them entirely, but roles involving routine information processing face genuine displacement pressure.
Can governments really control how AI reshapes the economy?
Yes, though the tools are familiar rather than new: tax policy, labour law, public investment in retraining. The question is political will, not technical feasibility.
