Necessary Conditions for Open-Ended Evolution Empirically Validated in Chromaria
L. B. Soros[1]
and Kenneth O. Stanley[2]
[1]Cross Labs, Cross Compass Ltd., Tokyo, Japan
[2]penAI, San Francisco, CA USA
lisa.soros@cross-compass.com
The quest for open-ended evolution (OEE) in artificial life centers on constructing artificial evolutionary processes that make new discoveries indefinitely as evolution appears to do in nature. Though this phenomenon has been a longstanding topic of interest in the alife community, the field generally lacks consensus on its exact definition (Juric, 1994; Bedau et al., 1998; Channon, 2003, 2006; Maley, 1999). For example, OEE has been described as the continual production of either novel (Bedau et al., 1997; Standish, 2003; Lehman and Stanley, 2011; Nellis, 2012) or adaptive (Bedau et al., 1998) forms. Recognizing that these descriptions are not necessarily mutally exclusive, OEE researchers have moved towards a pluralism that admits different degrees of openendedness (Taylor et al., 2016; Packard et al., 2019a,b). Nonetheless, the OEE community aims to replicate some dynamics of biological evolution (which is frequently interpreted as an effectively open-ended process) in the hope of creating more such open-ended processes.