In lending, it’s critical that any AI solutions are incredibly focused and meticulously checked. Especially for commercial loans with high-value transactions, the importance of the AI’s accuracy cannot be overstated.

 

The bottom line for lenders: AI should only be used for what it’s good at and where the reward outweighs any risks.

 

AI is incredibly useful when applied to the right areas. So many organizations seem to think that AI can do anything or fix any problem, but that’s a dangerous approach to take, especially for banks. AI cannot - and should not - do everything. Before implementing any solution, it’s important to know where AI can make a meaningful difference and where it only adds risk.

 

AI has its fair share of accuracy problems. It is known to make math or calculation errors. When it’s missing any information, it often fills in the blanks itself, not with facts but with information it has made up. It can also ignore complex instructions, again deciding for itself what it should do instead.

 

For banks, mistakes on a loan can be costly – especially high dollar commercial ones. If an employee was making these mistakes regularly, the bank likely would not keep them around for long. The same approach should apply to AI solutions. If a person lacks important skills, why would you hire them? If a solution makes certain mistakes, why would you implement it? Banks should not leave their high-dollar decisions in the hands of a machine, especially with its well-known accuracy issues.

 

Though AI is risky in certain areas, there are plenty of things it excels at. The real key to implementing AI solutions in a safe and sustainable way is by finding those areas where the bank has a need and where AI does well.

 

AI is great at finding or checking information. Lenders can spend a lot of their time checking certain requirements or referencing regulations, not because it’s hard work, but because finding the right information can be tedious. AI can cut those minutes or even hours spent down to seconds by surfacing insights more quickly, then letting the lender take over from there.

 

AI has also proven effective at writing and generating content, especially when it has a good bank of information to reference. This is another task that can take hours away from staff. Loan narratives can be incredibly lengthy, especially when the deal is large or complex. The bank already has all the information it needs, the time-consuming part is the writing itself. Letting AI compile all the information into the right format gives lenders more time to check that work, make decisions or interact with borrowers.

 

These are only a few examples of simple but time-consuming tasks and they are a perfect example of where the risk is low and reward is high for bankers. This is where the focus should be when considering a sustainable way to leverage AI.

 

Banks can automate most any task with AI, but it’s been proven time and again that some of those applications hurt more than they help. In reality, there are still plenty of tasks that need a human touch. Many times, the right balance is a blend of AI and humans working together on the same task.

 

Commercial lenders have a lot of tasks that are great candidates for AI and automation. Some of those tasks can run autonomously, but most of them need intervention from a banker. Implementing AI safely may look like letting AI create a first draft of a credit memo and letting a human take over and edit from there. Similarly, AI can quickly reference certain bank policies or local regulations, but it’s up to the human to interpret and apply those rules.

 

There’s not a world in which AI should be making credit decisions on large, bespoke commercial or agricultural loans. AI can have all the information, but it will never know a borrower like a human does. Community banks have built their business on personal relationships and knowing their customers. Using AI in the wrong place takes that away. Of course, using AI safely protects the bank from portfolio risk, but it also preserves the business model that makes community banks unique.

 

Implementing AI safely and sustainably is all about balancing risk and reward. Before going all in on AI, banks need to ask themselves: Where do I need help? What is AI good at? Where do those needs and capabilities overlap? And, most importantly, where should I keep control?

 

Lenders that approach AI the right way will see efficiency gains that help keep them competitive without exposing them to extra risk.