A system is said to be artificially intelligent if it does "smart" things, that are typically associated with humans. Conventional computer science consisted of creating algorithms that could solve problems unambiguously, in a step-by-step manner, like sorting a list. In AI, the tasks do not typically admit a step-by-step solution, such as face recognition, self-driving cars, etc.
This involves AI algorithms that learn patterns from data, to perform inference on new data. For example, given a lot of photographs labeled as of person A or not of person A, machine learning involves developing algorithms which when given a new photo can identify whether it is of person A or not.