ay Drew was reviewing a loan application from an online trip planning business last month and started to worry its service looked uncomfortably close to something an AI chatbot could do in minutes or seconds. He passed on the loan. As the founder of SBA Collective, a network of lenders and advisers focused on Small Business Administration-backed lending, Drew has been having discussions about the impact of artificial intelligence on small business since 2023. But now, the talk is more frequent and much more direct: He’s bringing it up with the borrowers and would-be borrowers of $5 billion-in-assets Winston Salem, N.C.-based Truliant Federal Credit, where he’s managing director of SBA lending.

“We have to at least ask the question, how is AI disrupting your business? There’s really not a single industry where you don’t need to ask that question,’’ says Drew.

Investors, economists and ordinary Americans are understandably obsessed with the pros and cons and the winners and losers from the advance of artificial intelligence. Much of the financial focus has been on the impact on workers, as big companies attribute layoffs (accurately or not) to AI, and on Silicon Valley, as investors try to figure out which software companies are themselves about to get disrupted. The $9 billion iShares Expanded Tech-Software Sector ETF has fallen some 20% this year as markets wrestle with the changes underway and on-the-way.

What has received far less attention is how the same uncertainty is playing out on Main Street, among accountants, consultants, law firms and other businesses, and the small banks that make decisions about financing them every day.

The stakes are high. Firms with fewer than 500 employees account for about 44% of U.S. GDP and employ 46% of the private workforce, or more than 62 million Americans. A good chunk of these small businesses operate in fields where generative AI appears most useful. A 2025 Microsoft research paper analyzing hundreds of thousands of AI interactions found the technology is, unsurprisingly, most applicable to tasks involving writing, research and analysis. One-third of small employer firms are in professional services and real estate (a single category), or business support services, according to the Federal Reserve’s Small Business Credit Survey.

Sure, most of the impact from AI may not be seen for years. But that’s little comfort to small business lenders who typically make loans with a ten-year life. While the Small Business Administration backstops about 75% of a 7(a) loan, banks still absorb losses if a borrower fails and must show the agency they followed proper underwriting standards to collect the guarantee. Over the past decade, roughly 8% of SBA loans have defaulted (on a cumulative basis), according to Lumos Data, a firm that analyzes SBA lending performance.

Small-business lending is as much art as it is a science. The science is in the spreadsheets. Bankers analyze cash flow, review collateral and check a potential borrower’s credit history. The art is the harder part: judging whether the business itself makes sense and has long term survival potential, given the high new business failure rate. (According to the Bureau of Labor Statistics, just under 35% of the businesses started in 2013 were still around a decade later.) Normally, the bankers can draw on their own experience; for example, did similar businesses survive the last recession?

But every so often something comes along that doesn’t resemble anything they’ve seen before, making forecasting harder than usual. A global pandemic. Tariffs that appear, disappear and appear again.

AI is today’s big known unknown. So far, lenders are not pulling back from entire industries or predicting a wave of business failures. But the way they think about risk is starting to change. If a company’s product is knowledge, advice or analysis (as opposed to, say, installing and servicing heating and cooling units), lenders have to at least consider what happens when software starts doing more of that work. Some of the borrowers banks thought they understood best are now harder for them to figure out.

The result: Bankers are increasingly demanding whether and how a small business loan applicant is thinking about and harnessing the disruptive impact of AI.

“Don’t try to sweep it under the rug,” advises Jeremy Gilpin, chairman of $280 million-in-assets Community Bankshares in LaGrange, Georgia. Firms that treat AI as a taboo topic raise concerns, he says. Lenders assume employees are already experimenting with these tools. If a company claims it is not using AI at all, that can signal something worse. Either employees are experimenting with the tools without management realizing it, suggesting a lack of oversight, or the business is falling behind a technology reshaping its industry.

Instead, Gilpin says borrowers should show how the technology fits into the business. If projections assume efficiency gains from AI, lenders want to see where those gains appear in the workflow. Policies around data security, quality control and employee use also come up more often. In some cases, lenders are asking to review those policies alongside the business plan.

Handled well, exposure to AI does not automatically hurt a credit application. In some cases it can strengthen one. Gilpin says lenders want to see owners who acknowledge the disruption and explain how their firms plan to adapt to it.

Legal work illustrates the tension between AI as a productivity tool and AI as a technology that could eventually replace some of the work lawyers bill for. Some attorneys already use AI tools to summarize case law or draft early versions of briefs. That can make firms more productive. It can also compress the hours traditionally billed to clients. Lenders reviewing those firms are trying to understand how the economics will evolve as those tools spread.

Plus, trying to sweep AI aside is unlikely to fool anyone. After all, small banks themselves are experimenting with the same tools and know how quickly they are improving.

Chris Hurn, founder and CEO of Lendesca, a fintech that builds AI-driven software to help banks originate and process SBA loans, says many underwriters are already testing large language models to help draft loan narratives and credit memos. Hurn says a credit memo that once took a week to produce can sometimes be drafted in “two hours or less.”

As AI takes over more of the mechanical work of underwriting, lenders may have more time for the other half of the job: the art of judging whether a business will still work in the future. “The more clerical the task, the more it has a bullseye on it,” Hurn says.

So far lenders say they have not seen loans fail because artificial intelligence wiped out a borrower’s business. But it’s still early. Most of the impact so far is showing up in deal selection and diligence rather than defaults.

For borrowers, the takeaway is that clean financials, cash flow and collateral still matter. But lenders increasingly want to understand how a business plans to operate in a world where software keeps getting better at doing what people used to do. In the end, that’s the art of lending.

© Forbes