The Central Bank of Kenya (CBK) last month announced on June 27 the licensing of an additional seven digital credit providers (DCPs), bringing the number of licensed DCPs to 58.
The regulator has continued to pay close attention to business models, consumer protection, fitness and propriety of proposed shareholders, directors, and the overall management of DCPs by ensuring compliance with relevant laws.
While there has been mphasis on regulatory compliance, challenges still remain. According to FinAccess Digital Credit Tracker Survey 2023, digital credit borrowers cited high interest rates and short repayment period as the main challenges they face while borrowing.
Conversely, while there is an appreciation of a steady growth in digital lending in Kenya, DCPs continue to grapple with loan defaults in a highly volatile business environment, making credit risk assessment critical for business survival.
As such, DCPs are experiencing a shift from the conventional credit scoring models to use of technological solutions.
AI tools are, for instance, being adopted in credit scoring though the analysis of vast amounts of data, including non-traditional data sources, to assess a borrower's creditworthiness more accurately with the aim of predicting a borrower's likelihood of repaying a loan.
The benefits of the use of AI tools in credit scoring include but are not limited to accelerated decision-making, promotion of financial inclusivity through ease of access to credit by individuals and businesses, and the adoption and implementation of a more streamlined lending process by DCPs which minimises risk and increases profit.
There is no doubt that AI is here to stay. How then should DCPs and the CBK embrace this shift?
Both must be alive to the threats posed by the use of AI tools in credit scoring such as the lack of transparency in machine learning models, the potential for biased decisions in assessing a borrower’s creditworthiness, the risk of use of poor data quality by DCPs in the training of reliable AI tools in credit scoring.
In as much as these threats exist, what can be done to ensure responsible use of AI by DCPs in credit scoring?
In the licensing of DCPs, the CBK must seek to establish whether an applicant seeking to obtain a DCP licence is applying the use of AI in its credit scoring model.
The CBK should consider adopting measures for responsible use of AI that applicants ought to satisfy in mitigating AI-related risks in credit scoring.
Some of these measures may include a demonstration of a DCP's ability to tackle issues around bias, provision of sufficient proof that the AI model the DCP intends to use or is using relies on high quality data in assessing a borrower’s credit worthiness and the mitigation measures a DCP has put in place to minimise risk.
Perhaps, it is time CBK issued a guidance note on the ethical use of AI by DCPs in Kenya.
The writer is a commercial lawyer practising at MNO Advocates LLP
