​In one of our use cases we have studied systems and processes where intelligence varies over time. It grows over time.

We wanted to revisit this topic because it is very important for managers to understand the complexity of running systems that evolve over time.

We selected a somewhat system engineering article. Here are some important learning you need to have.

A Manager should understand that in a systems that learns

over time:

1) The evolution of data over time is crucial, so need to know that storing data in the right amount is essential

2) There are critical aspects in upgrading the algorithms and you should know there is a risk to manage in doing this.

3) The trigger for when and how to re-train thee algorithms is likely to be a concerted decision between data scientists and a

business person that understand the when the system is not providing any longer optimal results. Or when the expected improved results from deploying a new algorithm outweigh the risk to make a new deployment .

4) Measuring systems performances over time is very important to inform when to change the algorithms in production. 

https://mlconf.com/blog/how-to-apply-continual-learning-to-your-machine-learning-models/