What holds a society together when work stops feeling dependable? The sharpest version of the AI-and-jobs debate often assumes a sudden replacement of human labor, followed by an equally sudden scramble for answers. The evidence in the background is less theatrical and more revealing. A 2024 study from MIT, MIT Sloan, and partner institutions examining computer-vision automation found that only about 23 percent of wages tied to vision-related tasks are currently economical for firms to automate.. That matters because social stability does not depend only on what machines can do. It depends on what firms can afford, what households can absorb, and how much time people have to adapt.
That slower timetable changes the central question. If labor markets are pressured by AI, the immediate problem is not only displacement but volatility: irregular hours, weaker bargaining power, benefits that disappear when work shifts, and households forced to make high-stakes decisions with no margin for error.
In that setting, stability looks less like a grand technological fix than a buffer. Cash support is one of the clearest examples. A German basic-income experiment followed 122 adults for three years as they received €1,200 per month; participants still worked 40 hours a week, matching the control group, while showing greater willingness to switch jobs or pursue education. Researchers involved in the project reported that participants did not show evidence of abandoning work despite the payments.” In the United States, a large guaranteed-income experiment run by Open Research provided $1,000 a month for three years to about 1,000 participants in Texas and Illinois, alongside a comparison group receiving smaller payments with a comparison group receiving far less In the United States, a large guaranteed-income experiment run by Open Research provided $1,000 a month for three years to about 1,000 participants in Texas and Illinois, alongside a comparison group receiving much smaller payments thdrawal from daily life but reinforcement of it: rent, food, transportation, debt, savings, childcare, medical visits, and the ability to help relatives. Some participants worked slightly less, but the time often shifted into caregiving, job search, training, or recovery from crisis. The payments did not eliminate hardship, but they did reduce the financial instability surrounding everyday expenses.
Public arguments about automation tend to revolve around meaning: whether people will feel purposeless if software takes over enough tasks. The references here point to something more practical. Before any philosophical rupture comes the administrative one. People lose equilibrium when bills arrive faster than income, when a better job is too risky to pursue, or when one extra shift threatens childcare or disability eligibility. The large U.S. experiment repeatedly surfaced the same lesson: people used modest financial slack to become more deliberate. They paid off arrears, improved credit, sought treatment, changed jobs, or made time for children. Everyday order came from predictability.
That also suggests a limit to the fantasy that either markets or AI alone will sort things out. Even the MIT study emphasized that lower deployment costs and AI-as-a-service models could speed adoption later, while higher computing, data, and labor costs could slow it. In other words, the pressure may build unevenly rather than all at once. A stable society in that environment relies on institutions capable of cushioning economic shocks and helping households adjust to changing work patterns…: wage supports, benefit systems that do not punish transitions, retraining that aligns with real demand, and cash assistance that treats households as capable decision-makers.
If AI weakens the old link between labor and security, everyday life remains stable through something more ordinary than futurist prophecy: enough time, enough cash, and enough institutional flexibility for people to keep their footing while the economy changes underneath them.
