Emerging challenges from molecular dynamics calculations of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mi>PbZrO</mml:mi> <mml:mn>3</mml:mn> </mml:msub> </mml:math> and their implications for the correct understanding of

Lead zirconate, ${\mathrm{PbZrO}}_{3}$ (PZO), is a perovskite crystal with an exceptional diversity of possible low-symmetry structures. Its phase transitions that are driven by temperature, field, pressure, and epitaxy are therefore of intense interest from experimental and theoretical viewpoints. Those viewpoints do not yet match in many important aspects. We conduct machine-learning-based molecular dynamics (MD) simulations to investigate several critical properties of PZO and identify recurring experiment--theory inconsistencies that often go underreported. We show that (i) MD correctly reproduces the experimental decoupling of antipolar displacements and octahedral rotations under pressure; (ii) MD predicts a positive electrocaloric effect, in contrast to experimentally observed negative values; and (iii) electric-field-induced transitions in MD proceed from $Pbam$ to $R3c$, whereas experiments suggest a transition to $R3m$. These findings reveal challenging limitations in current MD models---particularly in capturing entropy-driven phase stability---that must be addressed to reconcile simulations with experimental behavior in PZO.