Business leaders think AI-powered credit analytics is the next big thing
SINCE the advent of AI, the financial landscape has never been the same.
AI has been leveraged in many ways, such as virtual assistants or chatbots to enhance customer experience and predictive analytics to detect potential fraud.
These days, the use of AI in making credit decisions have been gaining traction. It is extremely attractive as it promises faster, more accurate decision-making for the lender.
As a result, an increasing number of businesses are betting on it said a recent World Economic Forum study. Thirty-eight percent, to be precise.
Undeniably, implementing AI in credit decision making is not easy. It is financially draining, takes a long time, and success is not guaranteed.
As such, this article aims to provide a more robust, up-to-date understanding of the utilization of AI in this manner, highlighting what business leaders think to be its benefits, challenges, and steps forward.
# 1 | What can be gained from AI-powered credit analytics?
A decrease in credit default is a significant driver for leaders to adopt AI, where about 15 percent expect to see credit default fall by more than 25 percent within two years of implementation.
Also, 89 percent of those surveyed drew their insights primarily from credit scores.
Interestingly, more are turning towards unconventional avenues as their go-to sources: over 50 percent of respondents said they are leveraging purchasing habits/POS and geo-location data.
# 2 | What is the main drawback?
When it comes to credit decisions, the risk of discrimination is high.
Take the American consumer lending market as an example. Research by UC Berkeley showed that compared to their counterparts, Latin/African Americans were charged a good 8 basis points more for the same mortgage products.
Most respondents recognize bias as a significant challenge in credit analytics. While AI has been shown to dial it down, skepticism still abounds.
About 15 percent of respondents believe AI will ‘backfire’, exacerbating discrimination in decision-making.
It is also worth noting that most of these respondents are users of non-traditional data. Psychometric testing was perceived to be the main avenue of bias (75 percent), followed by social media data (64 percent), and browsing preferences (60 percent). Credit scores were considered the least prone to increasing bias (46 percent).
# 3 | What’s the best way to leverage AI?
Those that wish to use AI must first evaluate their organization’s needs. Is there a need, and is their workforce ready for it?
If so, the next step would be to determine which data source works best. Not all data are created equal, and alternative data is not always better-the report shows that aside from psychometric testing (at 13 percent), the default reduction is insignificant between users of traditional and non-traditional data.
Finally, leaders must ensure that employees have the right mindset about AI-powered credit analytics. Get all this right, and business leaders can have full confidence in whatever decisions they make.