You will now "hack" the model. Computational models are only as good as the data coming in and the algorithms used on that data. Algorithms in code are written by humans, who by our very nature have biases, so that is one way they can sneak in. In our climate change example, we would need to look at the algorithms to see how they are calculating the change in variables over time. It was written by an educator, so we can assume it probably highlights variables she/he wants their students to understand, but it doesn't take into consideration many other variables that effect climate. Another way bias can creep into a model is from the data source. This model doesn't have a data source, but there are climate change models that use historical data. Historical data often includes biases because it represents the data deemed important by the collector, who was human and had biases. A recent example of this was uncovered by MIT researcher Joy Buolamwini who found that the training data sets used by many commercially available facial recognition software systems were estimated to be more than 75% male and more than 80% white and therefore they were failing to recognize darker-skinned complexions.