Since 1980, the Arctic surface has warmed nearly four times faster than the global mean. Enhanced warming in the Arctic relative to the global average is referred to as Arctic Amplification (AA). While AA is a robust feature in simulations of the Earth’s response to anthropogenic forcing, state-of-the-art climate models rarely reproduce the magnitude of AA seen in observations. This model-observational difference has led to concerns that models may not accurately capture the response or sensitivity of the Arctic climate to greenhouse gas emissions. Here, we use CMIP6 data to train and validate a machine learning algorithm to quantify the influence of internal variability in surface air temperature trends over both the Arctic and global domain. By applying this machine learning algorithm to observations, we find that internal variability increases the pace of warming in the Arctic and slows global warming in recent decades, which inflates AA since 1980 by 40% relative to the externally forced AA. Accounting for the role of internal variability reconciles the discrepancy between model simulated and observed AA.