SESK: Stuff Everyone Should Know · sesk.info/ai-mse-punishes-outliers

Which loss function punishes big errors, like outliers, more heavily: MSE or MAE?

Q: Which loss function punishes big errors, like outliers, more heavily: MSE or MAE?

Mean squared error (MSE), because it squares each error. For errors smaller than 1, squaring makes them even smaller.

Mean squared error (MSE), because it squares each error.

Mean absolute error (MAE)
They punish outliers equally
Mean squared error (MSE) ✔

Why: MSE squares each difference between prediction and label. An error of 10 counts as 100, so large misses dominate. MAE just averages the absolute differences.

Huh, didn't know that: For errors smaller than 1, squaring makes them even smaller.

The dad joke

MSE is that friend who never lets one mistake go. It squares it and brings it up forever.

Like this card?

Play the game

References

Answer: Mean squared error (MSE). MSE squares each difference between prediction and label. An error of 10 counts as 100, so large misses dominate. MAE just averages the absolute differences.

Google ML Crash Course: Linear regression, Loss developers.google.com/machine-learning/crash-course/linear-regression/loss
Choosing between MAE and MSE depends on the dataset and how you want the model to handle outliers, with MSE penalizing them more heavily.
Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?Artificial IntelligenceTuring predicted that by about 2000, judges would ID the machine at most what percent of the time?

That's stuff everyone should know.

Which loss function punishes big errors, like outliers, more heavily: MSE or MAE?

References

More stuff everyone should know