Latest thought provoking piece from Marc R Gagné MAPP Senior Privacy and Data Advocate, Cyber Intelligence and Director @ Gagne Legal. See more about Marc here. Image from Pixabay here.

The financial industry has long been riddled with inequalities. These inequalities are closely aligned with social inequalities, which show bias against women and minorities. Access to services, approval for loans, and availability of products that fit their needs have all been problems in the past for these two groups.

But now, along comes Artificial Intelligence (AI) to hopefully remove the human bias that’s plagued the financial industry. FinTech promises a revolution and in many sectors, it’s already begun. One area where this is most visible is in payment systems.

Payment systems affect the way consumers pay for the things they buy, but that’s not all. These systems also affect the way payment companies and banks make decisions about their customers. It’s also about access to financial products.

All these factors have changed dramatically in the last few years and much of the driving force is due to AI.

AI Has Transformed Life for Many People

Today’s modern digital payment systems have made possible a tiny revolution in commerce. Digitisation has improved life for large swathes of the population:

  • Enabling small businesses to sell to a global market
  • Increasing access to credit for underserved populations
  • Allowing the minority-owned corner store to procure and sell better goods with less hassle

Driving the revolution are fintech operations, who are more likely to use artificial intelligence to make their products work better. The knowledge and insight they gain from AI allow product developers to create products and services that are more completely in tune with what people want.

How Does AI Help in Payment Systems?

Customer Service

Customer service is one area where AI can improve services for digital payment companies like PayPal, Venmo, Google Wallet, or Apple Pay. Virtual assistants are helping people already, and of course, they’re driven by AI.

Payment Authorisations

Another area of change is the authorisation of transactions. AI has decreased the amount of time it takes to authorize a transaction to the point where it’s almost real-time.

Product Sales

AI also adds value in product sales for payment systems. The deep insights allowed by processing huge amounts of data allow companies to cluster their customers in new ways rather than relying on worn-out assumptions traditionally used by marketers. That can allow for more effective cross-selling and up-selling of financial products.

Collections & Debt Restructuring

When borrowers miss payments, two more aspects of payment systems comes into play: collections and debt restructuring. AI can improve these systems too, by providing more highly-customized solutions. When these solutions are more closely aligned with borrower’s needs and changing circumstances, everybody wins.

Customer Retention and Customer Satisfaction

Finally, when payment companies use AI to improve their customer retention processes, their churn rates tend to go down. AI helps them identify which customers they’re at risk of losing, and then helps them understand what they can do better in order to keep those customers.

Now for the Downsides

Data breaches make big headlines. Credit and debit card data gets compromised a little too regularly for most people’s tastes. Mobile payments are a huge concern for many users, as evidenced by a 2012 study by the U.S. Federal Reserve. That study revealed that data security was the reason given by 42 percent of consumers as to why they do not use mobile payments.

Another concern is privacy. Payments by nature involve multiple organisations, each a part of the payment ecosystem in its own important way. The U.S. Federal Trade Commission outlines several players who may be involved in any given transaction:

  • the merchant
  • a bank
  • the credit card processor
  • the developers of the software that runs everything
  • the manufacturers of the hardware used in the transactions
  • app developers
  • mobile phone companies
  • coupon companies
  • loyalty program administrators
  • the billing company for the mobile carrier
  • the broker for international transactions

Aside from the obvious privacy and data security concerns, there are sociological issues at play here, too.

AI Systems Tend to Have Bias

Just because AI is logical and bases its decisions on data, doesn’t mean there isn’t bias. The data we feed into AI systems can be skewed, representing the latent bias we all hold. Technology can and does perpetuate human bias.

Even the language we use is understood by AI to have certain biases. Some examples include associating “woman” and “girl” less often with mathematics and more often with the arts. Another example is associating African-American names less often with “pleasant” stimuli. The worry is that machine learning is reflecting traditional biases back at us.

Bias in the Payment Systems Industry

In fintech and specifically the payment systems industry, bias is more often than not against women and minorities. It may be because of traditional biases but there’s that other factor at play: in finance, there’s simply a larger volume of data on white men than any other segment of the population.

The financial industry has traditionally been largely the domain of white males. Likewise, the people who’ve traditionally been the investors are also white men.

Even though that’s changing, much of the data that exists on financial behaviour and outcomes come from that population. Women and minorities, newer to the finance industry as investors, borrowers, money managers, and household decision makers, are less represented in the data that AI systems feed upon.

That means any algorithm that’s charged with making decisions that benefit the largest number of people will skew its decisions to favour white men. Since it (wrongly) thinks women and minorities aren’t affected quite so much by its decisions, an AI system will minimize their importance in the algorithm it uses.

AI Bias is Hard to Manage

Another problem is that scientists often don’t even know how an AI system makes its decisions. That opaqueness can make it extremely difficult to pinpoint the inequality or bias. Even if technologists could find a way into the ‘black box’ of the AI decision-making process, they’d have a hard time studying the data because it’s protected. Financial data is, thank goodness, not part of the public domain.

AI, Bias, and Payment Systems: What Can be Done?

So, what to do? For now, one step is for companies simply recognize that the bias exists. That’s a tall order since it’s been a long haul just to increase awareness of data security and privacy issues in the fintech industry. We’re still grappling with data breaches and privacy concerns and the legislation isn’t fixed yet, either. Though they’re fighting the good fight, proponents for fixing bias in the system have a long road ahead of them.

Marc-Roger Gagné CCIE, CHTI, CCII, CCTA, CIPP/G/C, CTFI, MAPP

Services Juridiques Gagné Legal Services

@OttLegalRebels

Negligence, Arrogance and Thievery: How to View the Equifax Debacle


If you would like to have your company featured in the Irish Tech News Business Showcase, get in contact with us at [email protected] or on Twitter: @SimonCocking

Related Stories:
close-link