New Delhi, Dec 4 Machine learning (ML) is improving access to credit in India and reducing bad debts, as 93 per cent of lenders using it for vehicle loans report higher approvals, a report said on Thursday.
Machine learning is helping lenders improve portfolio performance and accelerate digital decisioning, according to the report by Experian which said that 90 per cent of people have reduced bad debt in credit cards using this tool.
The report based on feedback from 109 senior Indian credit decision makers, reveals that 79 per cent believe machine learning has expanded access to new customer segments and drove financial inclusion.
As much as 71 per cent of respondents said machine learning improved profitability by enhancing risk prediction and reducing bad debt, and close to 68 per cent cited improved risk‑prediction accuracy and operational efficiency as key benefits.
“ML is not just improving acceptance rates and reducing bad debt; it is helping build more transparent, efficient, and inclusive credit journeys,” said Manish Jain, Country Managing Director of Experian in India.
“As India continues its digital credit expansion, institutions that invest early in ML and GenAI will be positioned to compete, comply, and innovate," he added.
Lenders now confidently increase automation, with 71 per cent saying that ML allowed them to automate more credit decisions, reduce manual workloads and quickened decision making.
As many as 78 per cent believe most credit decisions will be fully automated within five years.
Generative AI is emerging as a powerful productivity tool in credit risk, with 84 per cent respondents saying it can cut time to develop and deploy new credit‑risk .
However, 65 per cent of non‑adopters said implementation costs outweigh the perceived benefits, while 44 per cent did not fully understand machine learning’s value, 54 per cent worried about model transparency and 55 per cent feared regulatory misalignment.
Disclaimer: This post has been auto-published from an agency feed without any modifications to the text and has not been reviewed by an editor