The augmented human-technology dilemma

Artificial Intelligence is gaining traction as many parts of the traditional banking process get automated with intelligent algorithms. Machine Learning is often seen as a different area of research in the field of AI, but realistically the entire machine intelligence world is exploding.

On Machine Learning: We might see supervised, unsupervised and re-enforcement learning as the current world, but new algorithms are created that bring big data, artificial intelligence and computational capacity into focus. This drive towards more accurate and predictive machine behaviors, leave us debating the role of humans and the interface with technology. We’re seeing that banking processes are automated and AI assisted by allowing for better ways to on-board customers, more predictive sales approaches, recommendations on transactions, credit score automation, fraud detection, behavior based marketing, etc.

I postulate that we can break this up into three stages namely; basic machine intelligence, response to learning, and augmented machine-human environments. Firstly, machine learning and other parts of artificial intelligence related fields have been implemented in various forms in banking including risk management, detecting fraudulent transactions, etc. All these approaches require the use of large data sets and complex algorithm training environments. Secondly, we’re seeing the focus shifting towards self-correcting and learning approaches where the purpose is to find a combination of algorithms that can behave like humans. This requires access to large, but more reliable, data sets, where the focus is on prediction and automated responses to effectively replace humans in key banking processes. As customers use more digital channels including mobile phones, we find that the need for certain roles disappear entirely in the banking service delivery channel. Thirdly, we are in pursuit of the most profitable interface between human and machine behavior. With a rapidly changing technology landscape comes the new world of redrawing boundaries of where humans are most likely to operate, opening the field of computationally enabled social science. Finding the optimal augmented view of work to be done requires an approach where different classes of algorithms can be used. Infusing these algorithms with human-error, a representation of social behavior, will push us into a new competitive AI ecosystem.

With these inventions moving multiple disciplines into automation, finding your place in this complex ecosystem will be an ongoing challenge.