Machine Philosophy - A Foundation of Philosophical Methodology

Abstract

This essay designs an evolved architecture for doing philosophy, which I call ‘machine philosophy’. I identify the core issues of current philosophical practice as an unhealthy mix of boolean argumentation, obsession with ordinary language, and a lack of methodological clarity. Machine philosophy entails that philosophical methodology should be continuous with that of the sciences. Specifically, philosophical theories are descriptive and objective, and the activity of philosophical theorising should be governed by the norms of statistical learning. In this regard, machine philosophy makes two core claims. One: intuitions are fallible evidence in philosophy, which reflect objective facts about our socio-linguistic realities. Two: philosophical theories are descriptive of our socio-linguistic realities in virtue of being statistically adequate models of our intuitions. This new architecture does not demand, but enables philosophical theorising to utilise formal, computational, machine learning mechanisms. However, unlike pluralistic proposals, this enabling of distinct mechanisms places a hierarchy on the epistemic quality of each method, measured by their ability to produce true descriptions.