Machine consciousness in Artificial intelligence - The Hard Problem as an Interface Problem

2017. October 06. 13:00 to 15:00
Builidng I B 210
Department of Telecommunications and Media Informatics

Előadó: Garrett Mindt (CEU)


The hard problem is the problem of why there is any experience associated with the physical processes occurring in our brains? Recently I have argued (Mindt, 2017) that certain information-theoretic theories of consciousness, specifically Integrated Information Theory (IIT), face a slightly different hard problem, since such theories claim that consciousness is the result of a specific type of information processing. The talk will be divided into two parts. First, I will discuss the hard problem of information in the context of IIT. This will involve explaining the objection to IIT developed in Mindt (2017) and what this means for the hard problem. I will then offer what I take to be a reinterpretation of the hard problem that makes it more scientifically useful to those working in the neuroscience of consciousness. This will involve turning the hard problem into the Interface Problem (Mindt & Montemayor, MS). The Interface Problem is a collection of related sub-problems concerning the epistemology of AI. Once this is done, I will discuss what implications the hard problem as an interface problem has on the project of machine consciousness in AI. The hope here will be to sketch out the epistemological tasks that lay ahead in developing the next generations of AI’s from a design-oriented perspective.

More information at the page of the Department of Telecommunications and Media Informatics.