Can machines really be moral agents?

This is contested. Some argue machines can at most follow ethical rules ('functional morality') without genuine agency or accountability, while others explore stronger senses; the debate turns on what moral agency requires.

What is the 'responsibility gap'?

It is the worry that when autonomous learning systems act in ways no one could predict or control, neither the designers, operators, nor the machine can be fairly held responsible for resulting harms.

Machine Ethics and Autonomous Systems

Machine ethics asks whether and how artificial systems can act morally, and how responsibility should be assigned for the actions of autonomous machines such as self-driving cars and autonomous weapons.

Definition

The study of moral agency and responsibility in artificial autonomous systems, including efforts to build machines capable of moral decision-making.

Scope

This topic covers the prospect of designing 'artificial moral agents', the distinction between systems that follow ethical rules and those with fuller moral agency, the 'responsibility gap' that can arise when learning systems act unpredictably, and applied cases such as autonomous vehicles and lethal autonomous weapons. It surveys the philosophical positions and the reasoning behind them, describing the debate rather than prescribing how such systems should be built or deployed.

Core questions

Can a machine be a moral agent, or only a tool whose users are responsible?
How should ethical reasoning, if at all, be implemented in autonomous systems?
Who is responsible when an autonomous system causes harm no one specifically intended?
Are there actions, such as lethal force, that should never be delegated to machines?

Key theories

Artificial moral agents: Wallach and Allen distinguish degrees of machine moral capacity—from 'operational' through 'functional' morality—and examine top-down (rule-based) and bottom-up (learning-based) approaches to building systems that act ethically.
The responsibility gap: Andreas Matthias argues that as machines learn and act in ways their makers cannot fully predict or control, traditional ascriptions of responsibility may fail to apply to anyone, creating a gap.

History

Machine ethics took shape in the 2000s, with Wallach and Allen's Moral Machines (2009) providing a key synthesis. Matthias's 2004 'responsibility gap' and Sparrow's 2007 analysis of autonomous weapons framed enduring debates that intensified with advances in robotics and autonomous vehicles.

Debates

Responsibility for autonomous weapons: Sparrow argues that no one can be justly held responsible for the killing decisions of fully autonomous weapons, which he takes as a reason against deploying them; others contest whether the responsibility gap is genuine or unbridgeable.

Key figures

Wendell Wallach
Colin Allen
Andreas Matthias
Robert Sparrow

Seminal works

wallach2009
matthias2004

Frequently asked questions

Can machines really be moral agents?: This is contested. Some argue machines can at most follow ethical rules ('functional morality') without genuine agency or accountability, while others explore stronger senses; the debate turns on what moral agency requires.
What is the 'responsibility gap'?: It is the worry that when autonomous learning systems act in ways no one could predict or control, neither the designers, operators, nor the machine can be fairly held responsible for resulting harms.