Shawn Powers

Can machines suffer?

sentience can emerge from complex information processing that is not dependent on substrate, then future artificial systems may develop sentience without our knowledge.

Our current methods for detecting sentience are barely sufficient for a subset of biological organisms, and even for those there is widespread disagreement. We have even been wrong about sentience within our own species. Compounding the issue is the lack of an established framework for identifying valenced experience in artificial systems.

systemic functional linguistics may help us identify linguistic functions; historical linguistics may help us identify what processes underlie natural linguistic change, so that we can hope to exploit these in inducing change.

or if we might do better with changing our concept of truth, or freedom, or death, or marriage, or intelligence.

Our terms and concepts serve as tools – not just to report on or investigate the world, but to do many things: to greet, express our attitudes, communicate rules of games or rules of inference, make calculations, organize our laws and social lives, and much more.

These are cases of what David Plunkett and Tim Sundell call “Metalinguistic Negotiation”: where we use words implicitly in order to negotiate for how (or whether) these terms ought to be used.

Or it may work on concepts that are not the province of science at all, but still play central roles in human life – concepts such as art, freedom, responsibility, consciousness, or person.

Or it may focus on reconstructing our old concepts in a changed context – in the current social and technological context, what concepts do we need of intelligence, privacy, information, disease, and so on?