A non-informative prior is a type of prior distribution used in Bayesian statistics that aims to have minimal influence on the posterior distribution, allowing the data to drive the conclusions. It is often used when there is little or no prior information available about the parameters of interest, effectively representing a state of ignorance.