Will AI decide your next credit card application?

Why you should trust CardRatings.com

At CardRatings.com we discuss the most up-to-date news and trends within the credit card space. Since we first pioneered the concept of online credit card reviews in 1998, our team of financial experts has provided comprehensive and unbiased credit card reviews for more than 175 cards, plus hundreds of additional resource articles to help educate everyday cardholders so they can feel more confident about their card choices. All our content is written and reviewed by industry experts. Though our content may occasionally contain references to products from our partners, we maintain strict editorial integrity and advertiser relationships and compensation never influences ratings, reviews or featured products. The difference between editorial content and advertising must always be clearly stated. Learn more.

Artificial intelligence (AI) has really burst onto the scene recently. It has suddenly gone from a long-discussed theory to a tool that’s actively used by a wide range of businesses. It might even decide your next credit card application.

Financial companies are testing various AI methods to see if AI can help them make faster and more accurate decisions about which applicants to approve and what terms to set. But what does this mean for consumers?

Credit card companies are turning to AI in part to help them better assess how much risk different consumers represent. This may help some consumers gain access to credit, while shutting others out. Knowing how AI might affect this process can help you be prepared for future credit applications.

Traditional credit risk assessment

A good place to start discussing the potential impact of AI on credit applications is to look at how those applications have been handled in the past.

Even without AI, credit card companies already use a great deal of data to assess how creditworthy each applicant is likely to be. This boils down to looking at what types of consumer behavior tends to correlate with good or bad payment records.

There are two major components to this:

• Devising statistical models that are used to calculate credit scores. These scores can be tailored to specific types of decisions.
• Refining decision rules for deciding on credit applications. Companies generally go beyond a simple credit score and look at additional details from an applicant’s credit history and financial situation that might be relevant to the size and type of credit they’re applying for.

What these components have in common is that they each rely on a tremendous amount of historical data – both about consumer behavior in general, and each applicant in particular.

AI and credit risk assessment

So where does AI come in?

The hope is that AI can make the risk assessment process more efficient. This could include:

• Improving credit score models
• Making decision rules for approving credit more accurate
• Broadening the type of data that accurate decisions can be based on
• Making decisions less reliant on lengthy historical data
• Speeding the decision-making process
• Allowing the process to become less labor-intensive

According to Experian, the application of AI techniques to credit decisions was found to result in a 35% decrease in non-performing loans.

That type of outcome helps explain why credit card companies are investing a tremendous amount of money in their pursuit of AI tools. Naturally, their ultimate aim is to make their businesses more profitable. For consumers though, there are potential benefits and drawbacks to the increased use of AI to make credit decisions.

 ➤  LEARN MORE:How fintech is changing how we think about and use financial services

Potential benefits for consumers

Here are some of the ways consumers may benefit from the use of AI in credit decisions:

• A broader pool of available data could give consumers more ways to establish credit. A long-standing problem is that you can’t establish a credit record until someone allows you to use credit. AI models could base credit evaluations based on a broader range of transactions – such as rental or utility payments – to allow people to establish creditworthiness even if they haven’t previously used credit.
• The ability to base conclusions on fewer data points could help consumers establish credit more quickly. AI has the potential to spot predictive relationships faster than traditional methods of data analysis. This could allow individuals to prove themselves to be reliable with less credit history.
• More automation could make the decision process faster. AI has greater ability to make even non-standard decisions automatically, reducing the need for human reviews that delay the process.
• Taking human judgement out of the process could eliminate a potential source of discrimination. Besides slowing the process down, historically there have been times when the human element has added personal biases to the approval process. Those biases aren’t always even conscious, but an AI model can be more objective.
• Pricing may be more fair. Bias doesn’t just affect which applicants are approved. Because it can cause some groups to be perceived as riskier, it may cause them to be charged more. One study by the University of California at Berkeley found that automated decision tools charged minority borrowers 40% less than face-to-face lenders. AI automation could make pricing more equitable.

Potential drawbacks for consumers

For all these potential benefits, consumers should be aware that an increased reliance on AI in credit decisions may have some drawbacks.

• Even AI is still largely dependent on the past. There’s an old saying in computer science: garbage in, garbage out. No matter how sophisticated the AI model, it is likely to be somewhat dependent on historical data. To the extent that available data is limited for some demographic groups or certain economic conditions, it may take longer to establish a sufficient history for AI to properly consider all the relevant factors.
• A more automated process may make it harder to explain decisions to applicants. Some experts are concerned that the more credit approval decisions get taken out of the hands of humans, the less able financial company representatives will be to explain the reasoning behind those decisions.
• AI models may perpetuate past biases. Lenders generally are alleged to have practiced widespread discrimination against certain types of applicants in the past. As a result, historical lending data reflects that discrimination. The concern is that automated models based on that data might reflect similar biases in future decisions.

The above are reasons why the financial industry needs to be careful about how it implements AI-based decision tools.

What consumers should look for

As AI becomes more common in credit decision-making, here are some things consumers should keep in mind:

• Discouraged applicants may want to try again. If you’ve been turned down for credit so often that you’ve given up trying, AI may give you a reason to try again. The use of non-traditional data in credit scores and decision models could well yield different results. This is especially true if you feel you’ve been discriminated against previously.
• You may be able to establish credit more quickly. AI is touted as being able to spot predictive patterns more rapidly than traditional models. This could mean it will take fewer financial transactions for a person to build a credit history that lenders are willing to trust.
• It always helps to know the rules of the game. As methods of determining creditworthiness evolve, you should stay informed about what factors are used to make decisions. The more you know about how AI views credit applicants, the better able you’ll be to take steps to qualify.
• Payment history is always likely to be a key factor. Your history of making payments on time has long been the single biggest determinant of your credit score. Even as models become more sophisticated, chances are they will still make a link between making payments on time in the past and doing so in the future.

Credit card companies wouldn’t be investing in AI if they didn’t think it could make their business more profitable. However, that enhanced profitability could also benefit consumers. To the extent AI helps credit card companies identify more qualified customers and makes the process of applying for credit faster and fairer, it could be a win-win.