Noisy reasoners: Errors of judgement in humans and AIs

Abstract

This paper examines reasoning under uncertainty in the case where the AI reasoning mechanism is itself subject to random error or noise in its own processes. The main result is a demonstration that systematic, directed biases naturally arise if there is random noise in a reasoning process that follows the normative rules of probability theory. A number of reliable errors in human reasoning under uncertainty can be explained as the consequence of these systematic biases due to noise. Since AI systems are subject to noise, we should expect to see the same biases and errors in AI reasoning systems based on probability theory. © 2012 Springer-Verlag.