By 2020, this level of bemusement will be reserved for banks who fail to embrace machine learning.

Machine learning is not technology from the future. From risk officers using algorithms to predict fraud, to marketing managers working out the best sales campaigns with recommendation engines, it has become an intrinsic part of everyday operations. Words such as "deep learning" and "neural networks," once reserved for the technical folks in research institutes, are now commonplace in financial institutions.

Machines can do it better

The reason is simple: Machines are better than humans in a very small, well-defined set of problems, and those problems are very frequent in finance. Being able to systematically recognize trends in large amounts of historical data means no longer relying on instinct and experience but instead relying on sophisticated models that give humans control over complexity.

The reason is simple: Machines are better than humans in a very small, well-defined set of problems, and those problems are very frequent in finance.

Another key part of this change has been the freefall in the price of computing power, as well as the emergence of cloud providers that allow access to supercomputers for those who need them. But perhaps more important has been the emergence of tech companies who specialize in bringing this potential to consumers, making it business-friendly and application-specific. These days, there are plenty of companies, such as Feedzai, Featurespace and CrowdProcess, that have made machine learning part of the everyday toolkit of finance professionals.

Predicting the future

The result of this change is that after decades of effort to get possibly a one-percent improvement, finance professionals are now capable of reaching double-digit improvements in their ability to predict fraud, credit defaults or any other objective. User-friendly applications can provide finance-centric interfaces to solve extremely complicated machine learning problems, with full supercomputing power being made available to the people who know the problems better: finance professionals.

Machine learning is here, and is here to stay. By 2020, banks will speak with surprised derision of the times in which machine learning was considered something from the future.