What's the truth? The truth is that you will need to ground your machine-learning algorithms. The truth is that Ground Truth will help you make sure your algorithm is accurate and not just a "truth" machine. The truth is that Ground Truth is a statistical term for checking the results of machine learning for accuracy against the real world. Ground truth is a term from meteorology to describe a reality check for machine learning algorithms. In other words, it's the real-world version of what the algorithm thinks it sees. The term is significant in the field of statistics and machine learning, where researchers use data sets humans gather people with biases, emotions, and limited time- to make decisions. A ground truth can help ensure the algorithm isn't just learning from partial data. If a program incorrectly identifies faces in photos as cats, for example, ground truth can tell us whether or not there are cats in those pictures (spoiler: there usually aren't). Ground truth is a term from meteorology for independent confirmation at a site for information obtained by remote sensing. One example is a storm spotter reporting a tornado out in the field that a meteorologist tracks on Doppler radar. A storm spotter can see and say whether or not there's a tornado. Still, without knowing about all the other factors that go into predicting when and where tornadoes will hit, it's hard to know if there's one on the ground or just some weird cloud formation and thanks to Doppler radar, meteorologists can get an idea of storms are moving and how strong they are based on how fast they're spinning (and the speed at which they're spinning). If they notice no wind in that part of the state, we may start looking out for tornadoes!