Despite the existence of several scientific and philosophical theories of the nature of consciousness, it is difficult to see how we can make progress on machine consciousness without some means of testing for consciousness in AIs. In short, we need to be able to "detect" conscious/subjective experience in a given AI system. In this paper, we present some behavior-based possibilities for testing for synthetic consciousness and discuss their potential limitations. The paper divides into several parts.
|Original language||English (US)|
|Journal||CEUR Workshop Proceedings|
|State||Published - Jan 1 2018|
|Event||2019 Papers of the Towards Conscious AI Systems Symposium, TOCAIS 2019 - Stanford, United States|
Duration: Mar 25 2019 → Mar 27 2019
All Science Journal Classification (ASJC) codes
- Computer Science(all)