How does a machine "learn" anyway? As humans, we learn from prior as well as new experiences, but this is not always a linear, rules-driven process. With artificial intelligence such as machine learning, the same case can be made: a machine learns from prior experiences, and also from how exceptions are handled by any human intervention. I discussed this in a blog on The Digitalist a few years back,
Black Box or White Box? Machine Learning for Finance and Risk Processes.
When considering artificial intelligence solutions such as machine learning for finance and planning processes, there are typically two camps.
- One camp embraces machine learning, and sees it as a way to automate repetitive tasks, where recommendations made by an application are a bit more cut-and-dry. freeing up time for more value-added activities.
- The other camp looks at AI with suspicion, because the "learning" that takes place is typically a black box, and there is a desire to understand how an application arrives at any recommendations that it makes, especially when it comes to planning and predictive tasks.
To look at Explainable AI, TruQua and SAP have authored a blog on the topic, with a three-part series of follow-up podcasts on the topic.
