AI presents opportunity for P2P firms
New artificial intelligence (AI) tools such as ChatGPT and Google Bard have been making headlines for showing how machines can write poetry and even pass law exams, but the machine learning (ML) technology also has plenty of practical uses in peer-to-peer lending.
Tomaso Aste, professor of complexity science and director of the UCL Centre for Blockchain Technologies suggests AI can improve credit risk models by predicting loan rejection and default risk using existing P2P platform datasets.
He co-authored a recent study based on a decade of LendingClub data that found AI can reduce the default risk of issued loans by as much as 70 per cent.
Read more: AI could price consumers out of credit market
“There is a great opportunity here,” he told Peer2Peer Finance News.
“It is not only the P2P sector but lending in general. Automation – reliable automation – can allow small or micro financing in a way previously impossible.”
P2P lenders in the UK are starting to get on board, using AI to improve their processes on both the borrower and investor sides of the business.
Read more: P2P marketplaces predicted to expand into new marketplaces and adopt AI
Kuflink has signalled that it is working on AI-focused products for due diligence, collections, compliance and risks. The property lending platform has said it is building algorithms to help predict future trends based on historical data.
Sharia finance-focused P2P lender Qardus uses ML for credit risk scoring, while Crowd2Fund’s automated investing tool SmartInvest incorporates AI and was built following talks with Murray Shanahan, professor of cognitive robotics at Imperial College as well as technology leaders from IBM and Microsoft.
Analysts from European platform Robo.cash have predicted that AI and ML will boost the growth of the P2P lending sector during 2023.
Robo.cash attributed this to P2P lenders being “increasingly aware of the benefits of using technologies and integrating them into their business.”
However, Brian Bartaby, chief executive of Proplend highlights that AI is only as good as the data lake it feeds from.
“Where it can be used for simple repetitive tasks, it may be useful but for something as important as underwriting, I doubt it,” he said. “Many lenders who have attempted auto-underwriting have failed spectacularly.”
Read more: Revealed: Secret talks underway to roll back financial promotion rules
