Can we imagine an ethical status of AI systems without consciousness?

Anna Strasser

Keywords: multidimensional spectrum of sociality, gradualist approaches, consciousness, ethical & social status

TL;DR: Can a radically gradualist approach call into question consciousness as a necessary prerequisite for ethical considerations?

Abstract:

Short abstract (word count: 283)

Debates about AI consciousness are marked by enormous indeterminacy. Still, answers to questions of whether AI systems could, should, or cannot have an ethical status are urgent. For humans and other living beings, phenomenological consciousness is one of the necessary conditions for ascribing ethical status. In this talk, I pose a provocative question, namely, whether consciousness is also a necessary condition for AI systems?
Drawing on gradualist approaches that, e.g., extent the set of possible candidates for moral agency to non-human animals, I ask whether a more radical gradualist approach could argue for a wider spectrum that might also overcome the sharp boundary between conscious and non-conscious entities.
To this end, I address the question of AI Systems’ potential social status. Assuming a multidimensional spectrum of social interactions, I conceptualize interactions between unequal partners as asymmetric quasi-social interactions that do not impose the same conditions on all participants. Thereby, I can avoid the dichotomy between mere tool use and social interactions. Utilizing a radical gradualist approach based on the idea of family resemblance, I propose a distinct set of conditions for AI systems. Conditions such as consciousness, sentience, and reflective rationality are questioned, and minimal forms of agency and intelligence (socio-cognitive abilities that enable coordination with interaction partners and instrumental rationality) are proposed as sufficient. This shows that agency and intelligence can be disassociated from consciousness.
Returning to the more far-reaching question concerning the potential ethical status, I will discuss challenges arising from the in-principle assumption of the necessity of consciousness, which may motivate us to use the idea of a multidimensional spectrum to characterize the potential quasi-ethical status of AI systems.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 11