'24/7 automation': Can AI robots really work without human intervention? - WION

TEXT START: While AI robots run 24/7 in highly structured 'lights-out' factories, they cannot operate entirely without humans. Lacking tactile intelligence, they struggle with unpredictability, meaning humans are still required for complex decision-making.

THE DISSECTION

This article is a piece of institutional transition management—a document designed to perform reassurance while simultaneously normalizing full automation as the obvious trajectory. It reads like a factory floor manager's pep talk: "Don't worry, your job isn't dead yet—just shrinking, degrading, and contingent."

The architecture of the piece is doing quiet ideological work: it celebrates 80-90% automation rates as a feature, treats human labor as a cost-center to be minimized, and frames human involvement as a temporary limitation of the technology rather than a structural feature to be eliminated.

THE CORE FALLACY

The article treats current robot limitations—tactile intelligence, improvisation under failure, adaptability to unfamiliar physical situations—as durable moats. They're not. These are known problems being aggressively attacked by the entire AI/robotics industry. Boston Dynamics, Figure AI, Tesla Optimus, and dozens of others are running trillion-parameter models against Dexterity Benchmarks right now. "Lacks skin-like sensory feedback" is a 2026 engineering challenge, not a permanent barrier. The lag is real but the direction is not in question.

HIDDEN ASSUMPTIONS

1. "Complex decision-making" is a stable category. It's not. AI is explicitly targeting this domain. Every week, frontier models advance on judgment, risk assessment, and strategic adaptation. The humans in this article are being assigned the tasks AI hasn't commoditized yet, not tasks that are inherently immune to commoditization.

2. Teleoperation is a stable category of employment. It isn't. Teleoperation is degraded labor—controlling machines from a distance, doing the work AI can't do, but without the physical presence, union leverage, or location-based bargaining power of traditional manufacturing work. As exception rates drop (systems fail less often, AI handles more edge cases), teleoperation becomes supervising automation rather than actively controlling it. The workload collapses toward surveillance, not skilled intervention.

3. "Autonomation" is an equilibrium state. The article presents 80-90% automation as the destination. It isn't. It's a waypoint. The trajectory is toward 95%, then 99%, then the entire cycle. The human role is being continuously compressed, not stabilized.

SOCIAL FUNCTION

Lullaby / Transition Management / Copium Distribution.

This article is designed to be shown to workers in manufacturing, logistics, and operations to delay labor resistance by 3-5 years. "Look, we're not replacing you—we're just making 90% of your work automated and assigning you to exceptions and maintenance. You're still valuable." The message is calibrated to prevent strikes, quiet quitting, or unionization responses to visible automation while the displacement proceeds.

THE VERDICT

The article inadvertently documents the displacement circuit in operation: AI handles the routine, humans are assigned to the exceptions, exception rates decline over time as AI improves, human workload compresses, human value proposition erodes. This is not a story about humans and AI sharing work. This is a story about humans being progressively squeezed out of the productive economy while being given just enough "complex decision-making" assignments to defer the reckoning.

The 80-90% automation rate is not a comfort. It's a progress report.