How can we validate theory-derived indicators of consciousness in Artificial Intelligence? (opens in new tab)

Butlin et al. clearly explained some of the key questions on artificial intelligence (AI) consciousness, drawing on existing theories revolving around computational functionalism [1]. They propose selecting a set of neuroscientific theories warranting sufficiently high credence, deriving indicators of consciousness (IoCs) from them, and then assessing whether they apply to particular AI systems. Mainly using behavioral measures, the IoC approach has been previously applied to animal conscious...