Om hvorfor vi fremdeles ikke benytter sosiale roboter i omsorg: innsikter fra antropologiske studier av språkbruk

摘要

We are surrounded by artificial speech agents (chatbots). There are also large expectations about embodying dialogue systems in social robots; a term for autonomous machines in a humanoid packaging, that offers opportunities for interaction through language. Actors in the Norwegian welfare state envision a future where these machines contribute to solving pressing issues, such as a lack of social contact in care services. These imaginaries of the future assume that robots possess, or will soon possess, communication systems for robust interaction with humans. However, despite technical progress, natural language processing in naturalistic contexts remains a significant challenge. Drawing on anthropological perspectives on language use, including insights from pragmatics, this article aims to shed light on why we cannot expect social robots to solve the problem of lacking social contact in eldercare in the foreseeable future. A critical analysis identifies the so-called ‘code model’ of language as an underlying problem, and sketches an alternative that is better aligned with the empirical realities of human interaction. The challenge of applying robots for social contact is illustrated with studies on communicative practices between people with dementia and their caregivers. Finally, I discuss some implications of these empirical issues for robots in care, and the critical value of anthropological insights.