Artificial intelligence and the paradigm shift in nanomechanics

Studies on the mechanics of nanomaterials, and nanostructures have driven significant discoveries and advances through multiple approaches, including experimental nanomechanics, computational nanomechanics, and size-dependent continuum theories. However, these approaches have inherent limitations related to measurement uncertainty, computational cost, and the calibration of material parameters. Recent developments in artificial intelligence (AI) have introduced new opportunities to overcome these limitations and reorganize the existing methodologies in nanomechanics. In experimental nanomechanics, machine learning (ML) techniques are increasingly employed for automated image analysis, real-time signal processing, noise reduction, and intelligent control of manipulation. In computational nanomechanics, AI enables the construction of surrogate models and machine learning interatomic potentials (MLIP) that significantly reduce the computational expense of atomistic simulations. The role of AI in size-dependent continuum mechanics was also examined, with an emphasis on the data-driven calibration of internal length-scale parameters, efficient solution strategies for non-local and strain-gradient theories, and the development of variable order nonlocal models. This study reviews recent developments, critically examines existing limitations, and delineates future directions for rigorous and physically consistent integration of AI into nanomechanics. Finally, we show how AI goes beyond being a computational tool to become a transformative force, enabling predictive, efficient, and physically grounded exploration of nanomechanics.