Prerelease selections from the upcoming paper (Peer Reviewed and Published as Part of BICA*AI Conference Proceedings 2020):
Abstract: The field of human psychology is relatively well known. It is a broad field; however, when we start creating sapient and sentient computer systems, we may not know how an AI’s psychology may or may not be. While the idea of ‘Artificial Psychology’ started in 1963 by Dan Curtis (Crowder), it has made little progress.
We intend to lay out a basis for approaching the study of the Psychology of AI. We will articulate the related assumptions and layout a foundation for approaching such psychology, generally both with non-human intelligent systems. Approaching such psychology has been a particular question for many (while working on various cognitive systems models. As those systems are tested, especially models that ‘experience’ emotions by design, it has become increasingly clear that we need a framework or basis to build a further analysis of such a system’ psychology.
What further complicates this is that even human psychology can be subjective, but what happens when AI turns out to be alien with multiple types of psychologies can field unto themselves. For example, we might have a field of study called the Psychology of ICOM (Independent Core Observer Model) or the Psychology of OpenCog (Goertzel). Humans are different and unrelated, except at the highest level, where we can generalize between the psychology of humanity, ICOM, and OpenCog systems. In this, every new AGI (Artificial General Intelligence), the system could require a whole field of study just for that system type. All of it would fall under this meta category of the Psychology of Artificial Intelligence. First, we start with assumptions of needed ideas/words required to frame our line of reasoning.
Conclusions
We define the Approach to Psychology as a meta-approach to building a study of the psychology of a given instance of implements of a given cognitive architecture. We believe that we should lean more towards calling it a meta field of Meta Psychology as this goes to the fact that it is about a meta-approach to psychologies in the broader sense. With this understanding, we can further break down how we approach the meta psychological approach to research artificial general intelligence and the psychology of ICOM and other systems. We can apply what we learn to the study of other cognitive architectures in such a way as to correlate and provide a framework of understanding and comparison. In studying an example like ICOM, the process of addressing the questions above and creating the infrastructure for the field to be applied to other types of AI systems and thus give us the new field of Meta Psychology.
by Aleksander Kelley, and M. Amon Twyman