The term test set generation for AI is primarily used in the areas of artificial intelligence, big data, smart data and automation. It describes an important step in the development of artificial intelligence, particularly in machine learning.
If companies want to use AI to recognise faces or errors in products, for example, they need data that they can use to train and test the AI. A test set is a specially compiled collection of data that is used to test how well the AI actually works. During test set generation for AI, this test data is selected and compiled in order to realistically measure the performance of the artificial intelligence.
Imagine a company wants to develop an AI that sorts tomatoes according to ripeness. For the AI to work reliably, it needs a test set consisting of many photos of tomatoes at different stages of ripeness. The generation of the test set ensures that the AI is put through its paces and subsequent incorrect decisions, such as categorising green tomatoes as ripe, are avoided. Without careful test set generation for AI, there is a risk that the AI will deliver incorrect results in everyday use.
Deutsch 
English (UK)