How is AI used for checking someone’s eligibility for a loan?

AI technologies can be used to predict how likely a person is to repay a loan or mortgage. To do this, the system processes data about a bank’s previous customers, including information about repayments. The AI technologies learn from patterns in the training data – they learn which features make it more or less likely that someone will repay a loan.

When a new customer applies for a loan, the AI system assesses a range of data about that person and compares it to the data it has learned from. It then predicts how likely the new customer will be to repay their debt.

Predictions have been in use in consumer finance, like banks and building societies, for a long time. What is different now is the increasing amount of data and computing power, which enables the use of complex statistics.

On one hand, this may be helpful to people who have previously not been able to get a loan. The use of specific data about them or about a wider range of circumstances, instead of generic or narrower data, may allow those with a low credit rating to be considered, if the training data shows that they are low risk.

On the other hand, the use of more data may introduce new risks. For example, AI systems use large amounts of data that contain historical discriminations or biases. These include assumptions about the significance of someone’s race, postcode or gender. As a result, their predictions can amplify various forms of injustice – for example, you may be less likely to get a loan because of where you live, although data injustice manifests definitely in different contexts.

As predictive technologies get more complex, it is more difficult to explain what data is being used and to explain how decisions are made. Not knowing how decisions are made can make it difficult to prove when discrimination or injustice happens. This can make it harder to get compensation when a machine-led process makes the wrong choice and causes harm. The amount of data used in AI technologies also raises questions about consent, what personal data is being used, and how and where the data is stored.

AI-led decision-making in consumer finance offers new possibilities. But many say that we need transparency, auditing and regulation to ensure that its use is both efficient and fair.