Artificial Intelligence, or AI (also referred to as “machine learning”), is here to stay. The financial industry is already inundated with companies offering solutions ranging from underwriting models that consider more than a thousand different data points on each loan application to pricing models that maximize profit while reducing risk to AI-driven audio, video, and communication tools.
It is a vast field to explore, and the use cases in lending are only just beginning to be discovered.
Likewise, regulatory scrutiny of AI and its uses is increasing by the day. The CFPB, for example, addressed black-box credit models in a Circular published in 2022 stating, in part, that:
- Federal consumer financial protection laws and adverse action requirements should be enforced regardless of the technology used by creditors.
- Creditors cannot justify noncompliance with ECOA based on the mere fact that the technology they use to evaluate credit applications is too complicated, too opaque in its decision-making, or too new.
Before a bank even begins to consider developing a specific solution or hiring a third party to provide it, there are some basic concepts to understand about both the benefits and the risks of utilizing AI in lending.
In this series of blog posts, I’ll break down a conversation I had with ChatGPT, a general-purpose “language learning model” AI that is representative of major advancements in AI technology.
What follows is a remarkable demonstration of how far this technology has come. I hope it will serve both to inform you on the specific issues around AI in lending and to illustrate how incredible and ubiquitous AI technology is going to be for all of us.
A few things to remember as you read this Q&A style discussion.
- The conversation is presented unedited with the exception of light formatting to improve readability.
- This conversation occurred in real time as if I was chatting with a human.
- ChatGPT is not a financial industry-specific AI. It can discuss virtually any topic in as much clarity and detail as it did AI in lending.
Starting the Discussion: What Do You Know About What I Need to Know?
Query:
I’d like to discuss, in a conversational style, the benefits and potential risks of AI used in lending. What questions and topics should I ask you about?
Response:
Sure, here are some questions and topics you could ask me about regarding AI in lending:
Benefits:
- How does AI improve the speed and efficiency of the loan approval process?
- Can you explain how AI can help improve risk assessment in lending?
- How does AI increase access to credit for underserved populations?
Risks:
- How does AI’s reliance on data impact its ability to make unbiased lending decisions?
- What steps can be taken to increase the transparency of AI lending decisions?
- Can you give an example of how AI could be misused in lending?
These are just a few examples, but feel free to ask me about any related topics that interest you.
The Benefits of AI in Lending
Query:
Let’s start with the benefits. In your own words, please address benefit #1.
Response:
Certainly! One of the main benefits of using AI in lending is that it can greatly improve the speed and efficiency of the loan approval process.
AI systems are capable of processing vast amounts of data in a matter of seconds or minutes, whereas human analysts might take hours or even days to review the same information. This speed and efficiency can help lenders make quicker loan approval decisions, allowing borrowers to access funding more quickly.
Additionally, automating certain lending processes with AI can reduce the time and resources required to complete these tasks, making the overall lending process more streamlined and efficient.
Query:
Can you provide specific examples of the loan approval process that can be improved with AI?
Response:
Sure! Here are some specific examples of how AI can improve the loan approval process:
- Customer onboarding: AI can quickly analyze customer information, such as employment and financial data, to determine loan eligibility and make a preliminary loan offer.
- Credit scoring: AI can analyze a wide range of data sources, including alternative data sources such as social media and mobile phone data, to generate a more comprehensive credit score for a borrower.
- Fraud detection: AI can quickly detect patterns in large amounts of data to identify and flag potential instances of fraud, allowing lenders to take proactive measures to prevent loan fraud.
- Loan underwriting: AI can automate the underwriting process by analyzing loan applications, credit reports, and other data to make informed lending decisions.
By automating these tasks, AI can help lenders make loan approval decisions more quickly and accurately, while reducing the time and resources required to complete these tasks.
Query:
Thank you. Returning to the list of benefits you mentioned, can you discuss benefit #2?
Response:
Of course! Benefit #2 of using AI in lending is that it can help improve risk assessment. AI systems can analyze a wider range of data, including non-traditional sources, to make more informed lending decisions and identify potential risks. This can help lenders make better decisions about who to lend to, how much to lend, and at what terms.
For example, AI can analyze a borrower’s spending patterns to better understand their financial situation, and can also use social media data to gain insights into a borrower’s behavior and financial stability. This information can then be used to generate a more comprehensive credit score, allowing lenders to make more informed lending decisions.
In addition to providing more comprehensive risk assessments, AI can also help lenders identify potential fraud and other risks more quickly and accurately, which can help to prevent loan losses. This can ultimately lead to lower default rates and improved financial performance for lenders.
Query:
And benefit #3?
Response:
Benefit #3 of using AI in lending is that it can increase access to credit for underserved populations. AI can help to expand access to credit for those who might not have a traditional credit history or who face other barriers to accessing credit, such as limited access to traditional banking services.
For example, AI can analyze alternative data sources, such as mobile phone data and utility payment history, to generate a more comprehensive credit score for borrowers without a traditional credit history. This can help lenders to make informed lending decisions and expand access to credit for borrowers who might otherwise be unable to access funding.
In addition, automating certain lending processes with AI can help to reduce operational costs and increase operational efficiency, making it possible for lenders to reach and serve a wider range of borrowers, including those in remote or underserved areas. By increasing access to credit for underserved populations, AI can help to promote financial inclusion and support economic growth.
Summary of the Benefits of AI in Lending
Clearly the use cases for AI in lending are more varied than just “better underwriting.”
In my conversation with ChatGPT on the topic so far, we covered:
- Improving speed and efficiency in the loan approval process
- Improving risk management in lending
- Increasing access to credit for underserved populations
And that’s still just beginning to scratch the surface.
As important – if not more so – is addressing the risks inherent in utilizing artificial intelligence in the lending process.
That will be the subject of my next blog post on the topic of AI in lending. Following that, I’ll share the conclusion of my conversation with ChatGPT in which we discuss recommendations and best practices for a bank considering the implementation of this powerful technology in its lending operations.
To follow along, be sure to subscribe to our blog updates below. Thank you for reading!
John is the Chief Technology Officer for Premier Insights, Inc. and Radiant Lending Technologies LLC.
