AI Jargon Buster
AI news and the language around it, simplified.
What is Overfitting?
Overfitting happens when an AI model learns its training data too perfectly, including all the random noise and irrelevant details, rather than learning the general patterns. Because it has essentially memorized the specific examples it was shown, it struggles to perform accurately when it encounters new, unseen data. Think of it like a student who memorizes the exact answers to every practice test question instead of actually learning the underlying concepts. They will score perfectly on the practice test but fail the real exam because they cannot apply their knowledge to new problems. In technical terms, the model has lost its ability to generalize, meaning it is too rigid to handle the subtle variations found in everyday business information.
Why this matters to you
In a workplace, an overfitted model is dangerous because it seems highly accurate during testing but fails unexpectedly when deployed. This leads to poor business decisions, such as a sales forecasting tool that works for last year's data but fails to predict current market trends. Recognizing this risk helps teams ensure their AI tools are flexible enough to handle the unpredictability of actual business operations, protecting the company from costly errors caused by rigid, memorized logic.
How you might hear this
We need to adjust our training parameters because the model is showing signs of overfitting and will likely struggle with our new customer demographics.
AI Jargon Buster
Search any AI term, explained in plain English.
Type a term below and search. You will be taken straight to the tool.
Related terms
See how your CV performs against the ATS algorithms that screen candidates before a human ever reads your application.
Try the CV Optimiser →How AI job displacement actually works, what it means for your career, and what to do about it. Written by someone who has been in recruitment for 25 years.
When the Ground Shifts →