The Economics of Evolutionary Programming

In this paper, we present an economic model of evolutionary computation technologies, highlighting how alternative propertization and liability mechanisms shape the balance between “exploration” and “exploitation” in evolutionary research. We consider two interrelated effects of evolutionary experimentation: (i) the positive externalities that occur when superior mutations enhance performance, benefitting all subsequent implementations of the algorithm; (ii) the negative externalities that arise when inferior mutations cause harm. Our analysis reveals that experimentation reaches the optimal stopping point only with a combined use of propertization and liability instruments. If intellectual property frameworks fail to propertize benefits from evolutionary processes, limiting liability may be a second-best means of encouraging experimentation.

Keywords: AI, evolutionary computation, genetic programming, liability, patents
JEL Codes: K32, K41, O31, D62