Investree: ‘Some Cases need both Artificial and Human Intelligence’
Fraud is a growing problem for financial services, including the domain of financial technology (fin-tech). Through P2P lending services, fintech provides benefits that can never be provided by conventional bank institutions. One of the benefits is connecting borrowers and lenders on an online platform so that almost all investments and loan processes are done online and relatively faster than conventional bank. No wonder the popularity of P2P lending has been skyrocketing for the past few years, especially in Indonesia.
However, enormous fraud potentials are lurking behind the benefits. Working with massive amounts of data sets in an instant time manner undeniably opens up more opportunities for fraud in P2P lending. Luckily, artificial intelligence (AI) is starting to be implemented in P2P lending.
AI implementation in the form of machine learning can evaluate massive amounts of data sets in a real-time manner, for example, assessing borrowers’ data for credit scoring. In terms of fraud prevention, as machine learning’s data evaluation can detect any anomaly in a pattern, it reduces human intervention in the process so that it narrows chances of fraud.
However, the implementation of AI does not necessarily eliminate the role of human beings. For some cases, pairing up AI with human intelligence is required and resulted in high accuracy of analysis. One of P2P lending players that have implemented AI is Investree, a pioneer of P2P lending marketplace in Indonesia that has been registered in OJK since 2017.
“We have implemented AI, but not in all processes and segments,” Said the Chief Risk Officer Amalia Safitri.
“There are some processes in some segments where AI implementation can be very effective. And there are ones where AI implementation still needs human intelligence,” she continued.
Investree provides several types of business loans which are Invoice Financing, Buyer Financing, Working Capital Term Loan, and Online Shop Loans. According to Amalia, business loans such as Working Capital Term Loan have combined human intelligence and AI, while retailers such as Buyers Financing has implemented AI without human interference.
In the lending process of Working Capital Term Loan, Investree uses two kinds of data, which are traditional and non-traditional (alternative data). Bank statement is a type of traditional data uploaded by borrowers. AI helps human analysts by analyzing the bank statement and producing information on patterns of finance-related behaviors.
However, once the pattern comes out and flags any anomaly, it needs a human analyst’s interpretation whether the data signify the occurrence of fraud or other factors before jumping to a conclusion. This way, the combination of artificial intelligence and human intelligence will produce a high level of accuracy.
“For Working Capital Term Loan, we still have to have a combination of AI and human intelligence. Borrowers receive a certain amount of money every month, but suddenly this month is not like in the past months. From AI’s analysis on bank statement, human analyst can see and interpret finance-related behaviors,” She concluded.