Labour
Reimagined

02 / 08 — Less toil

There is dignity in people. Must there be dignity in toil?

Courbet gave two ordinary road workers the scale and attention usually reserved for grander subjects. Their worn clothes, bent backs and heavy tools make the effort hard to miss. Painted in 1849, The Stone Breakers was destroyed in 1945; we know it through reproductions.

I wouldn’t want to preserve this way of working. A machine that spared those labourers years of pain would be welcome. We can respect the people doing the job without preserving the difficulty.

But they still need an income. If the machine saves someone’s back and costs them their livelihood, we’ve only solved part of their problem. I’d want the shorter working day to be something they could afford to enjoy.

If we remove the toil, how do we keep the security?

03 / 08 — Out of danger

Some work, we would gladly give away.

In The Rescue (1855), Millais wanted to honour the courage of firefighters. The urgency comes from both the children in danger and the person risking his life to save them.

If a reliable machine could enter a burning building and bring a child out, I would want it to. Protecting the rescuer matters too. The promise extends to dangerous inspections, disaster recovery and work in places people should never have to go.

That “if” carries a great deal. A convincing picture tells us nothing about a machine’s reliability. Who decides it is ready? Who can intervene? Who takes responsibility when it fails? Those are part of making the technology useful.

What would we need to know before trusting it with a life?

04 / 08 — Learning by doing

Who teaches the next pair of hands?

Georges de La Tour’s Saint Joseph the Carpenter, painted around 1642–44, brings Joseph and the Christ Child together in candlelight. This is an intimate religious and family scene. The robot pair borrows its warmth, although a picture that moves us tells us nothing about what a machine feels.

Assistance can help people learn. In one study of 5,172 customer-support agents, an AI assistant increased issues resolved per hour by 15% on average, with gains varying across workers and evidence of learning. That is a useful example, not a universal promise.

I’d want to know whether the beginner can do more for themselves afterwards. Help with a difficult task can teach us something; having it done for us may leave us no wiser. We still need practice, mistakes and someone to ask why.

Can we design assistance that leaves us more capable?

05 / 08 — Sharing the gains

A good harvest. For whom?

Bruegel’s The Harvesters (1565) has room for work, food, company and a nap. It is an idealised landscape, commissioned by a wealthy merchant, but its people are doing more than producing.

Automation could give us more of that room: shorter hours, better services, things that more people can afford. It could also concentrate income and power among the people who own the machines.

A bigger harvest doesn’t tell us who gets to eat. If a machine lets us produce the same amount in half the time, someone still has to decide what happens to the hours saved. Who owns it, who can negotiate and what rules we set will all matter.

What would you spend a productivity gain on: more things, or more time?

05 / 08 — Sharing the gains

And who is left picking up the scraps?

Millet’s The Gleaners (1857) shifts our attention to three women gathering the ears of grain left behind after the harvest. The Musée d’Orsay describes the contrast between their repetitive labour and the abundance in the distance.

What troubles me is how close the abundance is. The women can see the harvest, but they’re still collecting what’s left. Producing more would be little comfort without a greater claim on it. The rusty robots inherit that problem.

We should also be careful with predictions. The ILO’s 2025 assessment finds that one in four workers worldwide is in an occupation with some exposure to generative AI. Exposure means tasks could change; it is not a forecast that a quarter of jobs will disappear.

What would a fair share look like?

From the working archive: this image was not in the final twelve-work poster set. It joins the story here in the digital exhibition.

06 / 08 — Making & meaning

Who made this? And what makes it matter?

Velázquez’s The Spinners connects the labour of making cloth with the fable of Arachne and Athena, a contest over skill and authority. Its tapestry quotes a painting by Titian. Questions about imitation and artistic status have a long history.

The AI version gives us a much emptier room. I can enjoy the light and atmosphere while wondering where all that human activity has gone. It’s a pleasing image, but looking back at the original makes me less sure how much of its story has survived.

For this exhibition, AI models helped choose and describe paintings, generate images and write critiques. I steered them when they wandered off and chose what to keep. The tools opened up possibilities for me, but enjoying the results doesn’t settle questions about consent, credit or other people’s livelihoods.

When making an image becomes easy, what do we choose to value?

07 / 08 — A voice at work

What if the machine becomes the manager?

Lowry’s Going to Work (1943) shows people heading towards the Mather & Platt works in Manchester. In the reimagining, the crowd is mechanical. Yet people needn’t be replaced by robots for their working lives to become more tightly controlled.

Software can allocate shifts, set targets and assess performance. There are good reasons to want help organising all that. An OECD survey of more than 6,000 firms records managers’ perceived benefits alongside concerns about accountability, explainability and workers’ health.

But imagine being given a shift you can’t work, then finding there’s nobody you can ask to change it. I’d want the people doing the work to help shape the system, and to have someone with the authority to put things right when it makes a mistake.

Who can say no? Who can ask for a decision to be changed?