Climate change exposes people to conditions that are historically unusual but that will become increasingly common over time. What kind of weather do people think of as normal or unusual under these changing conditions? We use the volume of social media posts about weather to measure the remarkability of different temperatures and show that remarkability changes rapidly with repeated exposure to unusual temperatures. The reference point for normal conditions appears to be based on weather experienced between 2 and 8 y ago. This rapidly shifting normal baseline means warming noticed by the general public may not be clearly distinguishable from zero over the 21st century, with potential implications for both the acceptance of global warming and public pressure for mitigation policies.The changing global climate is producing increasingly unusual weather relative to preindustrial conditions. In an absolute sense, these changing conditions constitute direct evidence of anthropogenic climate change. However, human evaluation of weather as either normal or abnormal will also be influenced by a range of factors including expectations, memory limitations, and cognitive biases. Here we show that experience of weather in recent years—rather than longer historical periods—determines the climatic baseline against which current weather is evaluated, potentially obscuring public recognition of anthropogenic climate change. We employ variation in decadal trends in temperature at weekly and county resolution over the continental United States, combined with discussion of the weather drawn from over 2 billion social media posts. These data indicate that the remarkability of particular temperatures changes rapidly with repeated exposure. Using sentiment analysis tools, we provide evidence for a “boiling frog” effect: The declining noteworthiness of historically extreme temperatures is not accompanied by a decline in the negative sentiment that they induce, indicating that social normalization of extreme conditions rather than adaptation is driving these results. Using climate model projections we show that, despite large increases in absolute temperature, anomalies relative to our empirically estimated shifting baseline are small and not clearly distinguishable from zero throughout the 21st century.