can you also point to any references/link where you found these texts. It may help to get some context around these points. however i will try to respond based on my understanding of these components
1. Adaptive models works real time and learns as and when the responses are captured. so if you put adaptive models into production for the first time, as is, without any learning, they would'nt be effective. However you could train your adaptive models first before putting them into production to get better performance out from them for the initial stage.
2. In contrast to Adaptive models, Predictive models are not real time, ie they dont continue learning. the predictive models are created using already available historical data.