Technological progress could deliver cheaper services and greater prosperity. Whether people experience those gains depends on decisions being made now.
An unemployed graduate and a supermarket throwing away spoiled food appear to have little in common. Both expose a question that becomes harder as technology advances: who is responsible when promises of progress collide with everyday insecurity?
As Americans vote in midterm elections on November 3 and South Africa completes special voting before its November 4 municipal poll, that question deserves attention beyond the technology industry.
AI could help businesses produce more, improve scientific research and make expertise cheaper. But an economy can become more efficient without becoming more secure for everyone living in it.
South Africa illustrates the difficulty. Official unemployment among economically active people aged 15–34 stood at 47.4% in the second quarter of 2026. For someone still seeking a first job, a promise that AI will eventually create new occupations offers little immediate reassurance.
Across Germany, France and Italy, recent protests have reflected grievances over living costs, education, wages and public services. Some demonstrations involved violence, while many remained peaceful. Their causes cannot be reduced to technology. They reveal existing pressures that governments must manage alongside technological change.
The danger is a transition in which gains arrive on company balance sheets before they reach household budgets.
If AI reduces the need for junior administrative work, firms may become more productive while narrowing the routes through which young people gain experience. Training alone cannot resolve that problem unless employers also create opportunities to use it.
Yet mass unemployment is not an established outcome. The International Labour Organization estimates that one in four workers holds an occupation with some generative-AI exposure, with job transformation more likely than complete replacement. Exposure describes potential task changes, not inevitable dismissal.
Policy can influence the difference: whether smaller firms obtain affordable tools, whether workers participate in redesigning jobs and whether competition passes savings to customers.
Europe’s refrigeration attacks show another dimension of the same governance problem.
Remote controls help businesses manage equipment efficiently. Compromised controls can turn that convenience into physical losses. Forescout identified nearly 13,000 exposed refrigeration-related service observations worldwide, not an equivalent number of hacked stores. Its research describes how ordinary control functions can be misused to alter operating conditions.
There is no established evidence that AI caused the recent incidents. But Britain’s National Cyber Security Centre expects AI to make aspects of cyber intrusion more efficient through 2027, even as defenders gain new capabilities. Organisations unable to maintain their systems could fall further behind.
That makes maintenance, procurement and accountability central political questions.
A council buying an automated service needs staff capable of evaluating it, contracts requiring continued support and a workable fallback when it fails. A national government needs institutions capable of investigating harm without making compliance so costly that only dominant suppliers can compete.
Over the next few years, the consequential divide may be between societies that build this capacity and those that merely purchase more technology.
One path combines wider access, stronger public services and better protection during employment transitions. Another delivers pockets of extraordinary efficiency surrounded by insecure work, vulnerable infrastructure and decisions people cannot challenge.
Neither path is technologically predetermined. Elections influence budgets, enforcement and institutional priorities, although no ballot guarantees a particular outcome.
The test for governments is whether people gain practical control over the changes affecting them.
A citizen should be able to ask why an automated decision was made, who benefits from it and who must put matters right. A more intelligent economy will need institutions capable of answering.
