Matthias Michel

Rapid progress in artificial intelligence (AI) capabilities has drawn fresh attention to the prospect of consciousness in AI. There is an urgent need for rigorous methods to assess AI systems for consciousness, but significant uncertainty about relevant issues in consciousness science. We present a method for assessing AI systems for consciousness that involves exploring what follows from existing or future neuroscientific theories of consciousness. Indicators derived from such theories can be used to inform credences about whether particular AI systems are conscious. This method allows us to make meaningful progress because some influential theories of consciousness, notably including computational functionalist theories, have implications for AI that can be investigated empirically.

The question of which mental functions require consciousness has recently come to the forefront because of its relevance for investigating animal consciousness. Finding out that an animal can perform a function associated with consciousness would count as evidence that it has conscious states. I argue that most of the empirical research interpreted as showing that some functions are associated with consciousness fails to show this. Instead, it merely shows that the relevant functions falter when based on degraded sensory signals—which is unsurprising. This issue hinders empirical research on the functions associated with consciousness. I explain how consciousness research can do better when investigating the functions that require consciousness. Until the relevant research is properly carried out, the presence of functions supposedly associated with consciousness in non-human animals should not convince the skeptics one bit.

To study consciousness, scientists need to determine when participants are conscious and when they are not. They do so with consciousness detection procedures. A recurring skeptical argument against those procedures is that they cannot be calibrated: there is no way to make sure that detection outcomes are accurate. In this article, I address two main skeptical arguments purporting to show that consciousness scientists cannot calibrate detection procedures. I conclude that there is nothing wrong with calibration in consciousness science.

Consciousness is scientifically challenging to study because of its subjective aspect. This leads researchers to rely on report-based experimental paradigms in order to discover neural correlates of consciousness (NCCs). I argue that the reliance on reports has biased the search for NCCs, thus creating what I call 'methodological artefacts'. This paper has three main goals: first, describe the measurement problem in consciousness science and argue that this problem led to the emergence of methodological artefacts. Second, provide a critical assessment of the NCCs put forward by the global neuronal workspace theory. Third, provide the means of dissociating genuine NCCs from methodological artefacts.