> ## Content Index
> Fetch the complete content index at: https://johnwlittle.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The Growing Gap Between AI Practitioners and Holdouts
- URL: https://johnwlittle.com/the-growing-gap-between-ai-practitioners-and-holdouts/
- Published: 2026-10-25T14:00:00.000Z
- Updated: 2026-10-25T13:59:59.000Z
- Description: The risks of using AI are real. So are the costs of sitting it out as others develop skills and workflows that compound their advantage.
- Author: John Little
- Tags: AI, Technology

Spend enough time [experimenting and building with AI](https://johnwlittle.com/tag/ai/) right now, and you start to notice an uncomfortable reality: we are splitting into two completely different working worlds.

On one side are the self-taught practitioners. These aren't necessarily machine learning engineers or Silicon Valley insiders. They are analysts, researchers, writers, and operators who decided to treat these models like serious instruments rather than parlour tricks. Over the past couple of years, they have quietly moved past basic chat prompts. They are giving AI agents access to code repositories, using them as collection managers, automating data triage, and turning hours of manual synthesis into a fifteen-minute review cycle. They know the models hallucinate, they know where the guardrails fail, and they have built personal workflows to verify, constrain, and audit the output.

On the other side are the holdouts—people who still insist that AI isn't actually useful, or who refuse to touch it altogether.

Many of them tried a free chatbot once in 2023, watched it stumble over a basic fact or write a generic paragraph, and permanently wrote off the entire technology. Others treat refusal as a badge of honor, viewing any use of AI as an intellectual shortcut or a passing hype cycle that will eventually blow over.

To be fair, there is also a camp holding out for deeper reasons: they see the disruptive, ethical, and security implications of where this is all heading and want no part of it. That hesitation isn't irrational. The risks are very real. But standing on the sidelines out of principle is essentially a bet that the genie will eventually be put back in the bottle—and history is full of examples to the contrary. Transformative, dual-use technologies don't get un-invented just because they make us uncomfortable.

What makes the growing gap so striking isn't a lack of access. Most knowledge workers today are sitting in front of the exact same tools. The divide is entirely about curiosity and operating discipline.

The people getting extraordinary leverage out of AI didn't wait for a corporate training seminar or an official playbook. They taught themselves through trial and error. They figured out that maximizing AI isn't about asking a magic box for answers; it's about task decomposition, context management, and knowing how to review an agent's work the same way a senior editor or intelligence officer reviews a junior analyst's draft.

Meanwhile, the skeptics are either waiting for a flawless product that requires zero adjustment to how they work, or hoping the wave will somehow recede.

We have seen this movie before. In the late 1990s and early 2000s, there were professionals who dismissed the internet and mobile communications as noisy distractions—or feared what persistent connectivity would do to their industries—while their peers were busy rewiring how information moved across the globe. Eventually, the holdouts didn't lose an argument; they just found themselves operating at a fraction of the speed and reach of everyone else.

The same thing is happening now, only faster. The gap between someone who uses AI to pressure-test ideas, write scripts, and process massive volumes of information, and someone who refuses to engage with it, is no longer a philosophical disagreement. It is a compounding capability gap.

You don't have to buy into the breathless hype of the AI industry to recognize when a lever works. The tools are flawed, the security implications are real, and human judgment matters more than ever. But pretending the capability isn't there—or wishing it away—just leaves the field to the people who took the time to learn how to drive it.