NextFin News - Amazon workers are already being constantly surveilled, while hidden contractors in India and Costa Rica watch hours of video footage to train warehouse-monitoring systems. Heather Stewart’s Guardian column, published on June 14, argues that this is not a side effect of AI at work; it is part of the business model now being built as Elon Musk and a wider cohort of Silicon Valley billionaires push robotics and AI deeper into everyday life.
Stewart’s sharpest example is small but useful. D4YRL, a robot magician, was rejected by the Magic Circle last week because it was not sufficiently human. The performance was technically strong, but the group concluded that it did not connect with an audience’s emotions in the way a flesh-and-blood performer does. On the surface this looks like a quirky anecdote; the real issue is whether employers start treating technically acceptable output as a full substitute for judgment, empathy and the social role many jobs still perform.
This is not about anti-technology nostalgia — it is about where automation changes the terms of work. Stewart’s point is that AI framed as efficiency often shifts costs rather than removing them: more monitoring for frontline workers, more invisible labor for contractors, and more decision-making pushed into systems that are hard to challenge. What really changes is the value chain. Firms get lower visible labor costs and tighter control, but only by adding a layer of low-visibility human work to keep the machines accurate, compliant and commercially useful.
That creates clear winners and losers. Companies gain speed, throughput and data on worker behavior; the pressure falls on employees whose jobs are broken into measurable tasks, and on contractors in India and Costa Rica who do the manual review that makes “automation” function. The real trade-off is not humans versus machines. It is whether productivity gains come from removing dangerous or monotonous work, or from making labor more surveilled, more fragmented and easier to squeeze.
Stewart, a longtime Guardian business commentator, is explicit about the line she wants to draw. She is not saying AI will fail in the workplace; she is saying that success measured only by cost cuts, throughput or surveillance intensity misses the point of what a job is for. That matters because Musk’s camp and similar tech advocates often present automation as inevitable, when in practice adoption is a series of choices about task design, appeals, accountability and what employers decide should remain human-led. Whether that logic holds up depends on a simple fact: many businesses are not buying AI for novelty, but for pricing power over labor and tighter operational control.
There is a practical counterargument, and it is real. In warehouses, manufacturing and logistics, robots can remove workers from dangerous or monotonous tasks; in customer service, scheduling and compliance, software can save time and lower error rates. The math doesn’t add up yet, however, if companies claim a clean leap from labor to automation while relying on constant surveillance and unseen contractors to make the systems work. The risk nobody is talking about is that firms may discover too late that the backlash is not just moral or political, but economic: workers, unions, regulators and customers tend to react once the human cost becomes visible. Stewart’s column is one voice, not a settled verdict, but it lands on the central question boardrooms cannot avoid: which forms of monitoring should be off-limits even if they are technically possible. The Magic Circle’s refusal of a robot performer is just a symbol; the harder fact is that the boundaries are being written one workplace at a time.
Explore more exclusive insights at nextfin.ai.
