Scalability could trump complexity in machine learning debate

Risk USA: banks “on the precipice” of adopting more complex models, says Goldman exec

The debate over the use of more complex and hard-to-explain machine learning-based models to make customer-facing decisions is approaching a tipping point, say senior model risk executives – one that could ultimately extend to more heavily regulated activities such as lending.

Banks have long veered between deploying simpler machine learning techniques that can inform models such as logistic regression analyses, versus those whose computational shortcuts might yield faster results but defy easy