A Methodology for the Assessment of AI Consciousness
The research and philosophical communities currently lack a clear way to quantify, measure, and characterize the degree of consciousness in a mind or AI entity. This paper addresses that gap by providing a numerical measure of consciousness. Implicit in o
- PDF / 104,267 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 35 Downloads / 183 Views
Abstract. The research and philosophical communities currently lack a clear way to quantify, measure, and characterize the degree of consciousness in a mind or AI entity. This paper addresses that gap by providing a numerical measure of consciousness. Implicit in our approach is a definition of consciousness itself. Underlying this is our assumption that consciousness is not a single unified characteristic but a constellation of features, mental abilities, and thought patterns. Although some people may experience their own consciousness as a unified whole, we assume that consciousness is a multi-dimensional set of attributes, each of which can be present to differing degrees in a given mind. These attributes can be measured and therefore the degree of consciousness can be quantified with a number, much as IQ attempts to quantify human intelligence. Keywords: Consciousness Assessment
Self-awareness
Definition
Measurement
1 Methodology Any definition of consciousness today will be contentious. Consciousness is a subjective sensation and each thinking entity has a unique experience of consciousness which no one else can share. Nonetheless, a way to measure a thinking entity’s level of consciousness—for example, on a numeric scale from 0 to 100—is needed. Obviously this task is problematic and impossible to do with precision or accuracy, but the exercise is enlightening. Our goal is only to measure the degree of consciousness exhibited, without making any assumptions about its implementation. We say that if an agent exhibits the given set of externally observable behaviors listed below, then it is conscious by definition, and conversely, if it is conscious, then these behaviors will be present. Specific cognitive architectures, e.g., Global Workspace Theory [1–3], are candidates for evaluation using the methodology proposed here. We do not propose any testable theory of consciousness, only an approach to assessment and, by extension, a concrete definition thereof. This assessment consists of a series of questions. Use this questionnaire to evaluate the degree of consciousness of a person, AI system, or any other thinking entity. Answer the questions and then compute the score. There is no clear consensus on the definition of consciousness; this test reflects the author’s personal definition. Other researchers with differing definitions of © Springer International Publishing Switzerland 2016 B. Steunebrink et al. (Eds.): AGI 2016, LNAI 9782, pp. 305–313, 2016. DOI: 10.1007/978-3-319-41649-6_31
306
H.H. Porter III
consciousness may create their own evaluation methodologies, or weigh the questions here differently to reflect their definitions of consciousness.1
2 Instructions Answer each question using the following scale: 0 – NONE Not present at all 1 – SOME Present, but at a level far below human levels 2 – ALMOST Substantially present, but still at a sub-human level 3 – HUMAN Present at a level typical of a normal human 4 – SUPER-HUMAN Present to a degree that exceeds human ability After answering all questions
Data Loading...
