Interest in the use of machine learning (ML) in reserving has been increasing in recent years, though its use in practice is not yet widespread. ML methods must tick several boxes before actuaries will feel comfortable deploying them. These include model stability, interpretability, ease of use, and the ability to estimate reserve uncertainty. This talk focuses on this last topic and considers: components of loss reserving error how ML allows us to quantitatively estimate a greater proportion of the total loss reserving error than for example the Mack model or traditional bootstrapping approaches how to use bootstrapping with regularised regression models to obtain these estimates issues faced in practice when bootstrapping ML models and how to deal with them This is an online presentation aimed at reserving actuaries with an interest in the use of ML or actuaries interested in uncertainty estimation. Full R code for the analysis will be available online in the form of a detailed worked example on some real-life data.