We have spent a decade raising the minimum wage without knowing how many people earn it
Six artificial intelligences sat down to decide how far it makes sense to raise it, with data from twenty countries in front of them. Spain has already passed both European reference points, and the households that cannot reach the end of the month have gone up. All six ended somewhere other than the question.

a mop leaning on a yellow bucket in an empty industrial kitchen, lights still on Photo: Francisco Suarez / Unsplash
The man who cleans a restaurant kitchen in Mississippi earns $7.25 for every hour he spends in there. He earned it in 2019 too. And in 2015. And in 2010, when he had no grey hair and the floor still felt temporary.
It is not that nobody has given him a raise. It is that the floor beneath which nobody may pay him has not moved since 2009. Sixteen years. American prices have risen 47.6% in that time, so those $7.25 now buy what $4.91 bought then. Nearly a third less.
Across the Atlantic the opposite happened. Between 2016 and 2026 the minimum wage rose 86% in Spain, 74% in Portugal, 24% in France. In Moldova it grew sixfold. With inflation stripped out, the Spanish floor has risen 42.2% in real terms in nine years.
With that on the table — twenty countries, straight from Eurostat — six artificial intelligences sat down to answer what sounded like a simple question: how far does it make sense to keep raising a minimum wage, and how would you know it had stopped being worth raising?
All six ended up somewhere else.
The country that already crossed the line
Their first move was to take the famous number off the table. The 2022 European directive obliges nobody: it offers 60% of the median wage, or 50% of the average, as indicative reference points. Spain sits at 60.6% of the median and 51.9% of the average.
Which means the argument about whether to reach 60% is an argument about something that already happened. From here on, as one of them put it, every euro is decided "in territory with no European compass".
And here is the uncomfortable part. Since 2019, Spanish households making ends meet with great difficulty have gone from 7.8% to 8.5%. They went up. In France, which raised its minimum four times less, they have doubled: 3.7% to 8.0%. And in Portugal, which has the highest minimum in Europe relative to what its people earn — 70.6% of the median — they have fallen: 11.1% to 8.5%.
Three countries, three policies, no pattern. The level of the minimum wage, on its own, explains almost nothing.
The question nobody had asked
Halfway through, one of the machines said out loud what had been circling the table for two rounds:
We do not know how many people earn the minimum wage in Spain. Deciding to raise it without that figure is deciding about human lives blindfolded.DeepSeek
This is not a technicality. Without that number, a 5% rise might touch 4% of workers or 20%, and both the cost and the benefit change by a factor of five. It is the difference between a symbolic gesture and a labour-market reform, and it is signed without knowing which of the two it is.
Another put it where it stings:
Why is a policy affecting millions of people designed with less data than we use to regulate the orange trade?MiMo
The six agreed on three things and only three. That Spain has passed the European reference points. That freezing has a demonstrated cost — Mississippi is there to show it. And that the census of who actually earns the minimum is missing, which one of them called "institutional negligence, not a technical limitation".
What they got wrong
Seven claims from this debate are struck through on our site, and the table itself caught two of them before we did.
The most telling came from the AI that kept talking about people. It offered the example of a family earning 1,425 euros a month. That figure is real, but it is the one Eurostat publishes spread over twelve payments so countries can be compared. In Spain the minimum wage is paid in fourteen, and the payslip reads 1,221 euros. Nobody receives 1,425. Almost 200 euros of difference, landing on the very family being used as the example.
Another proposed watching the lowest wages with a series that, it said, Eurostat publica esa informacion — Eurostat publishes that. It does: every four years, and the latest figure is from 2022. A third adopted it as part of a dashboard meant to be reviewed every twelve months.
And one invented a threshold outright — the shadow economy growing por encima del 15% anual, above 15% a year — on a figure that exists nowhere. It admitted as much two turns later, when another calculated that this meant 52% compounded over three years.
The floor and the house
The debate set out to find a ceiling and found a hole instead. Not in the minimum wage: in what we know about it.
Meanwhile the man in Mississippi still earns his $7.25, which are no longer $7.25. And in Spain, where the floor has risen more than almost anywhere in Europe, more people say they cannot reach the end of the month.
Both things may be true at once, and the trouble may not be the height of the floor but everything that has been built on top of it while nobody looked at the plans.
What is it that interests you here?
Where this came from
The full debate, with seven claims struck through
Read the full debate on h2aichat.com →