AI Glossary / Governance, ethics and safety / Beginner

Robot ethics

Definition

Robot ethics examines the moral questions raised by designing, building, deploying, and interacting with robots. It considers safety, privacy, autonomy, responsibility, social relationships, and the distribution of benefits and harms. Because robots act in physical and social environments, ethical assessment must examine how they fit into the activities and relationships around them, not only their programmed task.

What it means

A robot can affect people through movement, sensing, communication, and changes to work or care arrangements. A machine that completes its narrow task successfully may still create problems for nearby people or shift responsibilities in ways the organization has not considered.

Robot ethics therefore includes choices made by developers, purchasers, operators, and institutions. It overlaps with machine ethics, which studies ethical reasoning and behavior within artificial systems, but is broader than programming a robot to follow rules.

Possible future robot moral status is also debated. Practical responsibilities toward affected people do not depend on settling that debate or assuming robots have human-like feelings.

How it works

Assessment starts by examining the robot's intended role and the setting in which it will operate. Identify people who use it, people it acts upon, and bystanders who may have little choice about encountering it.

Consider physical risks, collected information, opportunities for meaningful choice, and changes to social or organisational routines. Ask which responsibilities move when a robot takes over a task and whether anyone remains clearly accountable for the surrounding needs.

Engage affected people during design and trials, then use what is learned to change the deployment. Technical performance tests are necessary but cannot answer every question about acceptable use. A robot's behavior must be assessed in context, including foreseeable misunderstanding, failure, maintenance, and the arrangements for requesting assistance or stopping an unsuitable operation.

Three practical examples

1. A workplace introduces a mobile delivery robot and examines not only collision avoidance but also whether its routes obstruct accessible passageways or make staff work around it in unsafe ways.

2. A social robot records speech to provide an interactive service. The organization considers what users and nearby people understand about recording, retention, and who can access the information.

3. A facility uses robots for routine assistance and reviews whether staff responsibilities for noticing wider needs remain clear, rather than assuming that completing the automated task replaces every function of the previous human interaction.

Think of it this way

Think of adding a new machine to a shared workshop. Its own controls matter, but so do the people moving around it, the work it displaces, and the responsibilities that change when it becomes part of the routine.

The analogy is limited. Some robots interact socially, move through changing spaces, or appear to understand people in ways that influence expectations. Ethical review must therefore consider communication and relationships as well as physical operation, without treating a human-like appearance as proof of human-like capacities.

Worked example

Consider an original toy trial of a robot that carries materials across a shared building. It completes 90 of 100 trips without staff assistance, giving a 90% unassisted-completion rate. Across those trips, observers record 12 occasions when a person using a mobility aid has to wait because the robot occupies a narrow passage.

The completion rate and obstruction count describe different aspects of the trial. The 12 occasions are not necessarily 12 different people or 12 failed trips, and they should not be subtracted directly from the completion percentage.

An ethical assessment asks how long people waited, whether alternatives were accessible, who bore the burden, and whether routes or operating rules can be changed. It also asks whether bystanders had an effective way to raise concerns.

The figures alone cannot justify deployment. They show why a success measure centered on the robot's task can miss consequences experienced by people sharing its environment.

Illustrative case study

Fictional case: Hazel Centre introduces an information robot in its reception area. Visitors enjoy the interaction, but some assume that its friendly conversation is private and that it can resolve complex problems like a staff member.

The center reviews the robot's presentation, information handling, and escalation process. It makes the robot's role and limitations clearer, provides an obvious route to a person, and seeks feedback from visitors who find the interface difficult to use.

This invented case illustrates how social expectations raise ethical issues even when a robot is physically safe and technically reliable. The organization must assess what people understand about the interaction, beyond the intended design.

Common misconceptions and mistakes

Robot ethics is not reducible to fictional laws of robotics or a list of commands. Real settings contain uncertainty, conflicting interests, and responsibilities shared across organizations and people.

Avoid assuming that a robot's successful task performance establishes the ethical acceptability of its role. Likewise, a friendly voice or expressive face does not establish feelings, comprehension, or trustworthy judgment. Responsibility should not disappear into statements that the robot decided. The relevant human and institutional decisions still need examination, even when behavior involves complex automation.

Business use

Businesses procuring robots should ask how the proposed deployment affects workers, customers, visitors, and bystanders. Review the intended task alongside the wider routine it enters, including accessibility, privacy, support, and changes to responsibilities.

Use trials to investigate consequences rather than only demonstrate capability. Establish routes for feedback, assistance, and correction, and review the arrangement after updates or changes in use. Responsible deployment requires a workable social and organisational setting around the robot, supported by evidence that extends beyond a successful demonstration.

People also ask

Does robot ethics apply only to humanoid robots?

No. Industrial, mobile, service, and other robots can raise ethical questions regardless of whether they resemble people.

Is physical safety enough?

No. It is essential, but privacy, autonomy, accessibility, social expectations, and accountability can remain relevant even when immediate physical risks are controlled.

Must we decide whether robots have rights before assessing a deployment?

No. Practical obligations toward affected people and communities can be examined without resolving debates about future machine moral status.

Check your understanding

A robot completes its assigned trips reliably but repeatedly blocks an accessible route. Does reliable completion settle whether its use is acceptable?

Answer: No. The assessment must include the effects on people sharing the space and consider changes to routes, behavior, or deployment arrangements.

Related terms

• Ethics of artificial intelligence
• Machine ethics
• Autonomy
• Privacy
• Accountability
• Safety
• human-robot interaction
• responsible innovation
• accessibility

Learn next

Next, map a robot's wider setting. Identify one benefit, one possible burden on a bystander, and one responsibility that could become unclear when the robot takes over a task.

More in Governance, ethics and safety

• Risk set - A risk set is the collection of study subjects or units that are being observed and are eligible to experience a specified event at a particular time.
• Risk stratification - Risk stratification is the process of assigning cases to groups based on estimated risk of a specified outcome.
• Risk tolerance - Risk tolerance is the amount and type of uncertainty or potential harm that an organization or person is prepared to accept while pursuing an objective.
• Robust AI - Robust AI is AI designed and evaluated to keep functioning acceptably when inputs, operating conditions, or other relevant circumstances vary.
• Robustness - Robustness is the ability of a model or system to maintain acceptable behavior when specified conditions change or disturbances occur.
• Sampling bias - Sampling bias is systematic distortion arising when the process used to collect a sample gives an unrepresentative view of the population or task of interest.

Sources

Paula Boddington, primary author commentary on robotics safety and responsibility: https://repository.uwl.ac.uk/id/eprint/8282/ EPSRC, Framework for Responsible Research and Innovation: https://www.ukri.org/who-we-are/epsrc/our-policies-and-standards/framework-for-responsible-innovation/ The author's accepted manuscript was read through the Oxford repository, alongside EPSRC's framework. The examples are original; this entry makes no jurisdiction-specific legal determination.

Previous: Robot as a service Next: Robot learning