Modes of Modelling and Simulation in Artificial Intelligence

The aim of this research project is to provide a systematic and critical reconstruction of the various modes of modelling and simulation in Artificial Intelligence (AI). This reconstruction operates on a conceptual level as well as through empirical investigations into AI research practice. It will thereby help to explain both the heterogeneity and the success of AI.

In order to develop this systematic picture, a taxonomy is introduced that cuts across the traditional symbolic/embodied, behavioural/cognitive, and human/non-human dichotomies in AI. The alternative taxonomy is based on the coordinates of formal versus material analogies and theory-guided versus pre-theoretic models in science. These distinctions have parallels in the computational versus mimetic aspects and in the analytic versus exploratory types of computer simulation.

Under the working hypothesis that the dependence of models on available mathematical and computational resources is more profound in AI than in other sciences, and in continuation of a line of pragmaticist accounts of the role of models and simulations in science, it will be demonstrated how available simulational resources drove, and keep driving, the various developments of the AI research programme.