Low-fidelity models are non-inferior to high-fidelity ones, which may induce overconfidence, per literature cited in the review