Autonomous technologies are no longer confined to screens and data centres; they are increasingly embedded in the machines, vehicles and infrastructures that shape the physical world we move through every day. From self-driving cars and delivery drones to humanoid robots on factory floors and social robots in care homes, this new wave of “physical AI” is beginning to reorganise how work is done, how cities are designed and even how care is delivered. The question is not whether these systems will change the material fabric of daily life, but whose interests that transformation will ultimately serve.
Perhaps the most visible frontier is mobility, where autonomous vehicles promise to reshape how streets are used and who they are for. Driverless taxis already operate in cities such as San Francisco and Wuhan, while mining companies run vast fleets of autonomous trucks thousands of kilometres from human operators. Advocates argue that removing human error could cut road deaths, reduce congestion and lower emissions if fleets are shared and efficiently routed. Yet history suggests that new transport technologies often bring unintended consequences: more road space can fuel sprawl, and cheaper automated trips may increase total vehicle miles rather than reduce them.
Beyond transport, autonomous systems are quietly remaking production and infrastructure, from smart factories to energy networks. Manufacturers now pair advanced robotics with AI-driven sensing and control, allowing humanoid robots to take on complex assembly tasks and enabling round-the-clock operation. In extractive industries, self-driving haul trucks move ore across remote mine sites, and autonomous underwater vehicles inspect offshore platforms with minimal human presence. These deployments promise higher productivity and improved safety, particularly in hazardous environments, but they also accelerate the decoupling of physical output from local employment.
Public services are becoming a testbed for autonomy in less visible but socially significant ways. Local authorities and utilities use drones to inspect power lines and wind turbines, monitor wildfires and survey disaster zones that are too dangerous or remote for crews on the ground. In healthcare, pilot projects in mountainous regions have used autonomous aerial vehicles to deliver medicines and blood products, reframing access to care as a logistics problem solvable by software and sensors. Such applications can extend essential services to marginalised communities, but they also risk deepening dependency on proprietary platforms and fragile supply chains if not governed carefully.
As autonomy moves into homes, hospitals and streets, the boundary between social and technical systems becomes harder to draw. Social robots are being trialled as companions for older adults, including in a Singapore care home where a humanoid robot was reported to reduce loneliness and support staff. These systems do not just act in the world; they interpret and respond to human emotion, norms and expectations, raising difficult questions about consent, dignity and the outsourcing of care. At city scale, computer-vision systems that guide self-driving cars also reshape how urban space is mapped, lit and regulated, subtly privileging what machines can see and understand.
The politics of this autonomous turn are only beginning to come into focus. Proponents highlight contributions to the UN Sustainable Development Goals, from safer roads to lower emissions and more resilient infrastructure. Critics point to labour displacement, new forms of surveillance and the spectre of autonomous weapons, where the same sensing and decision-making capacities are applied to lethal force. Ensuring that autonomous technologies enhance rather than erode the public interest will require not just better engineering, but new institutional arrangements in which workers, citizens and affected communities have a real say over how these systems are designed, deployed and constrained. The physical world is being rewritten in code; the democratic task is to decide whose script it follows.