The Shopper Monetary Safety Bureau reiterated its skepticism of machine studying and predictive analytics in a good lending report issued Friday that reveals a rise in actions in opposition to redlining and discriminatory practices.
The CFPB mentioned monetary companies shall be more and more formed by predictive analytics, algorithms, and machine studying however that expertise also can reinforce “historic biases which have excluded too many People from alternatives.”
Within the report, Patrice Ficklin, the CFPB’s honest lending director, mentioned that whereas she is inspired by packages that may increase entry to credit score, she is skeptical of claims that superior algorithms are a “cure-all” that may remove bias in credit score underwriting and pricing. Her boss, CFPB Director Rohit Chopra, had beforehand warned firms about relying too closely on AI and machine studying in making lending choices.
Ficklin additionally famous a “legacy of structural discrimination” that extends to mortgage, auto, scholar, and different credit score markets for individuals of coloration, within the report.
“The mission of honest lending is to interrupt these discriminatory patterns and practices and to advertise entry to credit score to create fairer markets for all,” she mentioned.
Final 12 months, the CFPB issued 4 honest lending actions alleging violations of the Equal Credit score Alternative Act and referred two issues involving allegations of discrimination to the Division of Justice, in accordance with the 43-page report.
In 2020, beneath former appearing CFPB Director Dave Uejio, the company filed one lawsuit and issued 4 DOJ referrals.
The CFPB mentioned will probably be “sharpening its concentrate on digital redlining and algorithmic bias.” As extra Massive Tech corporations provide monetary companies and merchandise, the company mentioned will probably be working to establish rising dangers and to develop acceptable coverage responses.
Ficklin additionally mentioned the CFPB “will proceed to combat discrimination that manifests as unfair, misleading, or abusive acts and practices,” often called UDAAP violations.
“Redlining, pricing discrimination, and appraisal bias are vital obstacles to honest competitors within the mortgage market, impeding the power of a person borrower to get credit score on honest phrases, thus stifling progress in communities throughout the nation,” Ficklin wrote.