Philosophical Issues in Robot Design

Sophisticated robot behaviour would need to be responsive to various factors that correspond to concepts of significant philosophical interest. In this (philosophically modest) sense such robots would need to "employ" such concepts, as explained below, even if this falls short of full-blooded conceptual thinking. Most of these issues are discussed in the papers by John McCarthy linked from the page on Computational & Philosophical Issues, and see also the works listed in the References on Artificial Intelligence (including several by Maggie Boden, pictured here).

Free Will and Self-Consciousness

Artificial intelligences, especially if embodied in robots, raise evident questions about the nature of freedom and intentional action. Moreover any robot that reasons about what it can and cannot do will have to employ some relevant concepts, including perhaps a concept of self (especially if it is to interact and communicate with other robots).

Language, Classification, and Speech Acts

Communication between robots will involve some sort of language, and unless this is entirely inflexible and "hard wired", the relevant terms and concepts will need to be capable of evolutionary development through usage and adaptation to circumstances. This will involve consideration of how observed objects are to be classified into novel "kinds"; imprecision and vagueness must inevitably accompany such linguistic evolution, forcing these issues also to be faced in a practical context. Other complexities will arise where the communication is intended to mediate coordination or to be adaptably purposive, involving consideration of speech acts.

Representation of Cognitive States

Any robot that has to operate in a context of uncertainty, and to interact with other cognitively fallible robots, would benefit from being able to represent a range of cognitive states, both of itself and others. Moreover such states would demand practical treatment even if they fall short – perhaps in numerous ways – of the full-blooded cognition enjoyed by humans.

Referring Expressions

One particular variety of cognitive uncertainty involves the application of referring expressions, the same sorts of considerations that motivated Frege's distinction between Sense and Reference and Russell's theory of definite descriptions. Practical ways of handling such uncertainties would be required to facilitate communication in an environment where robots' perspectives differ and questions of identity are significant.

Causation and Conditionals

Reasoning about actions and events will require suitable handling of causation, and also – if the system is to learn from experience – of counterfactual propositions. Its handling of both must be such as to be practically applicable (a constraint which arguably rules out some prominent contenders in the literature).

Robot Ethics

Any sufficiently sophisticated robot that is designed to act within ethical limits will require some ethical decision procedure, specific enough to yield practical results. This serious issue has only recently become widely recognised, as robots become closer to widespread use in contexts that are potentially life-saving (e.g. medical care and rescue from hazardous environments) or potentially life-threatening (e.g. self-driving vehicles and autonomous weapons).

Maggie Boden

Maggie Boden

Pioneer, philosopher and historian of the comparative study of human and artificial intelligence