Recently, there has been remarkable progress towards the development of large-scale quantum devices through advances in quantum science and technology. This progress opens new doors for proof-of-principle demonstrations of quantum simulations as well as practically useful applications, such as quantum-enhanced metrology. However, these applications require first quantifying how well a quantum device produces a desired target state, which is currently experimental challenging as existing methods for the quantitative verification of a quantum device require fine-tuned control and substantial experimental resources. In this talk, I will present a simple and efficient benchmarking protocol to estimate the fidelity of large-scale quantum devices. Our protocol relies only on time evolution of a quantum system undergoing Hamiltonian dynamics, followed by simple measurements without any sophisticated control and readout. Fundamentally, this simplification stems from a universal phenomenon associated with many-body chaos from generic, strongly interacting quantum systems. We demonstrate our benchmarking protocol experimentally for an analog quantum simulator based on a Rydberg atom array, and numerically for other quantum platforms such as superconducting qubits, trapped ions, and itinerant particles in optical lattices.