Main Street is racing to fit artificial intelligence into its business models, but most small businesses are still learning to walk before they can run.

Small business owners have already made up their mind about AI, no longer asking if they should use the technology, but rather how and when to integrate it. More than three-quarters of small business owners are already using AI to some degree, and more than 90% say it’s working, according to a survey of small business operators published by Goldman Sachs on Tuesday.

Higher efficiency and productivity are the primary selling points of AI for small businesses, and nearly seven in ten expect the technology to help grow their revenue. But while many entrepreneurs are experimenting with AI, a huge gap remains between enthusiasts and owners who are able to fully integrate AI.

The Goldman Sachs survey found that only 14% of owners have embedded AI across their core operations, meaning most small businesses have downloaded the app, but few have read the manual.

The survey polled 1,256 participants in Goldman Sachs’ small business education program between January and February, finding that many entrepreneurs are excited about AI but uncertain how to fully cash in on it. Barriers include a lack of technical expertise, difficulty navigating a crowded tools landscape, and data privacy concerns. More than 70% of respondents said they’d benefit from more training and implementation resources.

High risk, maybe high reward

Small businesses are warming rapidly to AI tools, but struggling to absorb them. A 2025 report from the Chamber of Commerce found 58% of small businesses used generative AI in their operations, more than double the share from 2023. AI tools including ChatGPT, Claude and Gemini vaulted social media to become the second most popular technology among small businesses, trailing only search engines.

Most small businesses use AI for everything from copywriting and content creation to automating customer service, with some even using the technology for complex tasks such as coding and website design.

But the more revenue-driving applications remain limited to a smaller share of tech-savvy firms. Less than a quarter of businesses surveyed by the Chamber of Commerce use AI to accomplish tasks that would likely translate to greater revenue growth, such as optimizing supply chains, identifying potential customers, and producing new insights on products and services. Distrust of AI remains rife among some small businesses, with data privacy a key concern for half of firms using AI, according to Goldman Sachs.

Another hurdle may lie with the customers themselves. AI currently ranks as one of the least enjoyable topics of conversation in the U.S., with one recent NBC poll finding that only 26% of Americans saw AI positively, and 46% had a negative view. Companies may also hesitate to deploy AI for client-facing tasks given general fatigue about the technology. A Gartner survey published this week found that half of consumers prefer their business going to AI-less brands, with many customers particularly abhorring the reliability of AI-assisted shopping experiences.

One bright spot: jobs. The wave of corporate layoffs and fears of AI-driven job loss fears sweeping white-collar industries does not seem to have affected small businesses yet, with the Chamber’s report finding that 82% of small firms using AI were able to grow their workforce over the past year.

But just as corporate America struggles to figure out how best to adopt the technology, small businesses might remain hesitant. Generative AI pilots at large companies have largely stalled so far and failed to generate significant revenue, and small businesses with fewer resources might be hesitant to take the risk on thousands of dollars’ worth in enterprise models and staff training.

A bill passed in the House last year, called the “AI for Main Street Act,” aims to allocate more resources to increase AI literacy among small businesses. But until larger companies prove that AI can be worth the investment, smaller firms may be reluctant to follow.