Experian survey: 44% of lenders say Machine Learning models ‘difficult to deploy’

Over 40% of lenders have identified Machine Learning models as a challenge to deploy effectively, according to new research by Experian on the future of digital lending.

The survey, conducted in January this year, sought responses from a mix of industry players, including banks, fintechs, the automotive sector and consultants, to gauge how far along on the digital lending journey the ecosystem is and the challenges that remain in adopting automation and the decisioning process.

When asked whether they found it a technical challenge to deploy modern Machine Learning models into production in their organisations, 44% of respondents said it is difficult to deploy Machine Learning models effectively.

The survey also sought views on the effectiveness of an organisation’s model regulatory and governance process, with 34% of respondents saying that it “works, but has challenges”.

However, more than half of those surveyed said their model was “very effective”.

In answer to a question about how they plan on adopting automation, AI and machine learning models, 35% of respondents said they intended to build in-house AI and machine learning models.

Nearly the same percentage, at 34%, said they would outsource automation, AI and machine learning to a new or existing partner.

During an Open Banking Expo live panel debate ‘The future of digital lending and enablers of growth’ in December last year, in association with Experian, speakers discussed customers’ expectations of fintech lenders.

Allica Bank’s chief risk officer Aileen Gillan said: “Ultimately, for the fintechs to succeed, we need to offer a better customer experience than the established brands.”

Simon Forster, a senior consultant at Experian, opened the panel debate by observing the increase in financial services products “being taken through digital channels”.  

Asked whether lenders are prioritising affordability, fraud prevention or eligibility in the current environment, Jason Maude, Starling Bank’s chief technology advocate, said that lenders “really have to go for all of these things at once, prioritising one over the other doesn’t really work”.