The AI Rabbit Hole You Can't Afford

Learn the basics of AI, then stop. You are too expensive to be a beginner. Learn enough to put AI to work on real tasks in your business, and let the learning stop there. The expertise that made you successful is the asset. Open-ended AI study is what puts it at risk.

That is Jeff Fargo's recommendation. The survey figures below are attributed, and our interpretation is marked as ours.

The learning budget you never counted

Owners of very small businesses already run out of hours. UENI's May 2024 survey of 837 US businesses with five or fewer employees found a median work week of 60 hours, with 62% of owners working more than 50 hours.

The same survey found owners dedicate an average of 11.2% of their work time to learning new things, and that 88% set aside time for professional development. That figure covers general learning, not AI study, and UENI did not measure how the time splits.

Our read is that AI study comes out of that same budget, because the hours in a 60-hour week are already assigned. For scale, 11.2% of 60 hours is roughly 6.7 hours. That is our arithmetic on a mean share against a median week, not a figure UENI published, and it describes microbusinesses in 2024.

Comfort is high. Confidence is not.

Adoption is no longer the question. Thryv's 2026 AI and Small Business Adoption Survey, conducted in April 2026 among 561 US small and mid-sized business owners and decision-makers who are not Thryv customers, put AI adoption at 66%, up from 55% a year earlier.

In the same survey, 86% said they were somewhat to extremely comfortable using AI. And 70% said they need more, or significantly more, training to use it effectively.

Most of these owners report being comfortable with the tools and still report wanting more training on them. The survey records both numbers without explaining why they coexist. Our read is that comfort and confidence answer different questions, and the gap keeps people hunting for a missing piece.

Where the pull comes from

The Thryv survey does name where owners learn. 57% said their primary training source is YouTube and social media. 49% rely on online resources and webinars. A third turn to AI tools like ChatGPT to ask how to use AI.

What we notice about those channels is structural rather than measured. A feed, a webinar funnel and a chat window are open-ended by design. They can always serve more. A course has a last page.

The risk we see is drift. A webinar leads to a newsletter. The newsletter names six tools. Each tool has a tutorial. Nothing in that chain tells you when you are finished, so the decision to stop has to come from you.

Enough: the working threshold we use

This is Fargo Factor's operating rule. No survey we found establishes a correct amount of AI knowledge for an owner, or shows that professional development is wasted.

Enough is a working threshold, not a certificate. You have reached it when you can get a useful result on a real task in your business, repeat that result on purpose, and recognize when the output is wrong.

The third part carries the weight. Knowing when a tool is confidently wrong is what makes the first two safe to rely on.

Fargo Factor has published the basics that clear this bar. Treating AI like a search box wastes most of what it can do, covered in Stop Using AI as a Search Engine. Prompt folklore does not improve output, covered in Saying Please to ChatGPT Won't Get You Better Answers. A tool that agrees with everything you say is not giving you information, covered in Stop AI From Telling You What You Want to Hear.

Short is the point. A threshold you can reach beats a syllabus you abandon.

What staying a beginner costs

"You are too expensive to be a beginner." Jeff Fargo, founder of Fargo Factor, Fargo Walks episode 103

Beginner time is expensive time, because it is spent at your rate and produces nothing a client pays for.

An attorney reading a sixth prompt-engineering thread is an attorney not practicing law. A contractor comparing note-taking tools is a contractor not bidding work. That is the trade we are pointing at, and it is an argument rather than a measurement.

What Thryv reports here comes from AI users describing their own experience. 92% of AI users said the technology saves them time. 79% said they expect to get back between 11 and 60 hours per month, which is an expectation rather than a recovered amount. 81% said AI has made them more strategic in how they run the business.

Why the stakes are rising

Owners are starting to weigh AI against headcount. In the Thryv survey, 46% said that if AI and an employee could perform the same task equally well, they would choose AI, up from 38% the year before. 13% said they hired fewer people because of AI, and a third said AI lets them do more with fewer employees.

That 46% rests on its condition. The question assumed equal performance on the same task, so it is a preference under a hypothetical, not a report of what owners did.

The direction still matters. Owners in this sample are treating AI as a capability question, and capability is decided by whether the work comes out the other side. That is a reason to get something working rather than to keep studying.

What this does not mean

This does not mean skip the basics. Learn them, and learn them properly. An owner who refuses to touch AI is making a worse version of the same mistake.

It does not mean AI is unimportant. Adoption at 66% among owners who are not that vendor's customers is not a trend to sit out.

It does not mean stop permanently, and neither survey shows that stopping is safer than continuing. That judgment depends on your trade, your margins and what you cannot currently do.

The limit is on open-ended learning. Not on using the tool, and not on your curiosity.

The decision

Name the two or three tasks in your business where AI would save you real hours this month. Pick ones you can describe in a sentence.

Learn only what those tasks require. Use the working threshold as your test: a useful result, repeatable on purpose, and you can tell when it is wrong.

Then stop, and go back to the work you are already expert at.

You are already the expert. Give yourself grace.

Set the review date

Stopping works only if the stop has an end date. Otherwise you have not stopped, you have drifted.

Put a recurring review on the calendar. We suggest quarterly. No research establishes a correct interval, so treat that as a starting point.

At each review, ask one question. Is there a task I would pay real money to hand off that AI could not handle last quarter? If yes, learn that one thing. If no, close the tab and go back to work.

That converts an open channel into a closed loop with a date on it.

Common questions

How much AI does a business owner actually need to learn?

No survey answers this. Fargo Factor's working rule is enough to get a repeatable, useful result on real tasks and to recognize a wrong answer. That is a short list of working habits rather than a course of study.

Is it risky to stop learning AI?

It depends on your business, and neither survey settles it. Scheduled reviews are how we manage that risk.

Why do owners who feel comfortable with AI still want more training?

Thryv reports both: 86% comfortable, 70% wanting more training. It does not explain the gap. Our read is that comfort and confidence measure different things.

How do I know when I have learned enough?

You can produce the result on purpose, twice, and you can spot the output that is wrong.

Sources

• Thryv, 2026 AI and Small Business Adoption Survey, issued via Business Wire, June 2026. Conducted April 2026 among 561 US small and mid-sized business owners and decision-makers who are not Thryv customers, through an independent third-party panel. The release does not name the research firm.
• UENI Research, How Microbusiness Owners Spend Their Time, May 2024. 837 US businesses with five or fewer employees, fieldwork May 6 to 13, 2024.

Both are vendor-published with partially disclosed methodology. Weigh the sample sizes and dates accordingly. The 6.7-hour figure is our arithmetic, not a figure UENI published.

I'm Jeff Fargo, founder of Fargo Factor. I help businesses streamline and scale with AI. https://www.fargofactor.ai