COVID-19 has had impact on how we collectively approach doing business in the financial space. We’ve seen substantial changes come as a result of the pandemic. The interest thing, though, is that not everyone can agree on what those changes were, and whether they were positive or negative developments.
As reported by Finextra, the Bank of England recently conducted a survey of financial institutions to gauge their impression of COVID-19’s impact on machine learning technology. The results were surprising; 35% of respondents said that covid-19 had a net “positive” impact on their machine learning and data science technologies. Specifically, respondents said this was in reference to how those technologies are used to support remote workers.
At the same time, another 35% of respondents said that COVID-19 had a net “negative” impact on machine learning model performance. How can opinions be so divided?
Strengths & Vulnerabilities of Machine Learning
As sited in the report:
“[Machine learning] models may perform poorly when applied to a situation they have not encountered before in the training data. This is particularly relevant in the context of the COVID pandemic when the underlying data may have changed or the statistical properties of the data may have changed.”
This passage highlights one of the key shortcomings of predictive technologies like machine learning.
Machine learning and data science are based on careful analysis of historical trends and information. Technologies based on these approaches are well-adapted to making predictions based on expansive data sets. However, this may render predictive technologies unsuited to act reliably in the event that circumstances suddenly change.
The pandemic produced a sudden and unexpected downturn resulting from shutdowns and ongoing concerns about subsequent waves of infections. These couldn’t have been forecasted on the basis of economic data or historical predictors, because the COVID-19 pandemic had no close historical analogue.
Lessons to Learn
This begs the question: where does machine learning technology go from here? What lessons can we take from the experience of the COVID-19 outbreak to help us refine these technologies and be more prepared to meet unanticipated factors that will come up in the future?
The most important point that we need to remember is that total reliance on machine learning was never the correct approach. No automated process, no matter how intelligent, will ever be flawless.
It’s been understood for years that machine learning’s reliance on historical data is a weakness, just as much as it is an asset. Take fraud prevention, for instance; new fraud threats appear every day. Fraudsters constantly seek to develop new strategies to attack cardholders, merchants, and financial institutions. They operate faster than machines can identify and respond even with the benefit of predictive analysis.
It’s true that machine learning technology carries tremendous potential for the payments and finance space. However, that technology must be just one element of a broader strategy. Human expertise, for instance, is a critical component that will be necessary to augment machine learning.
Augment Machine Learning With Human Expertise
We must leverage machine learning technologies in a way that compliments their inherent benefits. Computers can apply complex algorithmic logic to a situation much faster than a human. At the same time, humans think in a dynamic, multi-dimensional manner.
Computers operate incredibly efficiently in terms of linear problem solving. In contrast, humans think slower, but in a manner that is more akin to a hypothetical “quantum” computer. Each mode of operating carries benefits that can augment the shortcomings of the other.
At Fi911, we are proponents of deploying machine learning technology for the payments space with the careful oversight of human expertise.
Our Dispute Lab™ technology removes redundancy and makes resolving disputes more efficient than ever. FIs and payment resellers can receive, interpret, distribute, and process every cycle of a dispute—from retrievals to representments and arbitration cases—using queue-based routing and hundreds of customizable rules. This is all backed by some of the industry’s leading minds in the payments and fintech space.
Machine learning represents an incredible leap forward for payments. However, the COVID-19 outbreak serves to highlight the need to identify ways to deploy machine learning technology in the most effective manner possible. Schedule your below to learn more about how the cutting-edge approach to fintech deployed by the experts at Fi911 is helping financial institutions around the globe to protect transactions and prevent loss.
