The Question Nobody Wants to Ask Out Loud
It usually comes up sideways. An owner mentions that one employee has become genuinely good at using AI — drafting faster, catching errors, handling more volume — while others are still doing things the slow way. And then, quietly: Should the first person be making more money?
It's a fair question. Not a comfortable one, but fair. Compensation is supposed to reflect the value someone brings. If AI fluency is increasing someone's output, ignoring that entirely seems like a mistake. But rewarding people for using AI tools when the whole point of AI is to make things easier also feels a bit off.
Here's what owners are actually weighing, where this tends to go wrong, and the simpler path most land on.
The Case For Tying Pay to AI Skill
Let's take the argument seriously first, because there is one.
AI fluency — real fluency, not just having a ChatGPT account — is a skill. A good prompt engineer (someone who knows how to get consistently useful output from an AI tool) can produce in two hours what takes a less-skilled colleague a full day. A team member who can build and maintain a simple AI workflow, spot when the output is wrong, and adapt when the tool changes is genuinely more valuable than one who can't.
You already pay people more for skills that make them more productive: experience, technical knowledge, language ability, specialized certifications. AI fluency is another item on that list. If you don't acknowledge it in compensation at all, you risk one of two things: your best-at-AI people quietly notice they're being paid the same as everyone else, or you start having to pay a premium to hire AI-fluent people from outside rather than growing them internally.
The strongest version of the case isn't about rewarding AI use — it's about rewarding results that AI made possible. If someone's throughput has measurably increased, that's real value regardless of how they got there.
Where It Goes Wrong
Most owners who try to formalize "AI bonuses" or "AI skill pay" run into at least one of the following problems.
You end up rewarding activity, not outcomes. If you pay more for "demonstrating AI use," employees will demonstrate AI use. They'll add AI-generated summaries to emails that didn't need summarizing. They'll run reports through an AI tool before printing them out unchanged. They'll talk about AI constantly and produce nothing particularly different. You haven't bought better results — you've bought better-looking process.
You penalize people whose roles don't fit AI well. Not every job in a 30-person company translates to an AI productivity boost. Your most reliable warehouse supervisor, your best technician, your sharpest salesperson who closes deals through relationships — if they can't "show AI use" the way a marketer or account manager can, a formal AI skill premium will land on them as a pay cut relative to peers. That's a morale problem and a retention problem rolled into one.
It's very hard to measure fairly. How do you assess AI fluency? A quick quiz? A portfolio of outputs? Manager observation? All of these are gameable or subjective. The moment you try to systematize it, you'll spend more management time on the assessment than you'll ever save from the skill differential.
It creates perverse incentives around transparency. Employees who are great at using AI but modest about it won't self-promote. Employees who are mediocre at everything but good at looking like AI power-users will. That's not a hiring or culture problem you want to introduce deliberately.
What Actually Works Better
The more you dig into this, the more the compensation question turns out to be the wrong frame. The right frame is: are you measuring the outputs that AI should be improving, and does your compensation reflect those outputs?
If a marketing coordinator is now producing three times as many content drafts per week, the question isn't "should we pay her extra for using AI?" — it's "is her output, her quality, and her responsibility appropriate to what she's earning?" If someone has leveled up their role because they figured out how to use AI well, the conversation is a role conversation and a results conversation, not an AI conversation.
That reframing leads most owners to one of a few cleaner approaches:
Fold it into existing performance criteria. If output volume, quality, and initiative are already part of how you evaluate people, AI-driven improvements flow naturally into those ratings without creating a separate track. You're reviewing what changed, not how.
Treat AI fluency as a hiring and onboarding factor going forward, not a retrofit. New job descriptions can include expected proficiency with AI tools where relevant. New hires can be screened for it. That's cleaner than trying to restructure compensation for your existing team around a skill they didn't know they were supposed to develop.
Invest in enabling people, not just rewarding the self-starters. The employees who figured out AI on their own are probably going to be fine regardless. The ones who need training to get there are the ones your compensation system risks quietly penalizing. If you haven't given your team a fair shot at learning these tools — through time, resources, or actual guidance — tying pay to AI skill before doing that is putting the cart before the horse. The article on getting your employees to actually adopt AI tools covers this well: most adoption failures come from under-investment in enablement, not from lack of motivation.
Have the conversation directly. If someone is clearly more productive because of AI and you haven't acknowledged it, say so. "You've figured out how to get a lot more done, and I want your compensation to reflect that" is a direct, honest, human conversation. It doesn't require a compensation policy redesign.
The Key Person Risk Angle
One thing that does warrant genuine compensation attention: the employee who has become the unofficial AI lead for your whole team. They built the workflows, they know which tools do what, they trained everyone else informally.
If that person leaves, you're not just down a person — you're down institutional knowledge that nobody else fully holds. This is the AI-era version of key person risk, and what happens when the person who set up your AI quits is a more concrete version of a conversation many owners need to have. If someone holds that much organizational value, that deserves to be reflected in their compensation and in deliberate cross-training — not because they use AI, but because of what they've built.
The Short Answer
Paying employees more because they use AI is probably the wrong mechanism. Paying employees more because their output, responsibility, or role has grown — and AI is part of why — is just normal compensation management done well.
The instinct to recognize AI fluency isn't wrong. But formalizing it into a separate pay track creates more problems than it solves for most businesses at this size. The cleaner path is to get your performance metrics right, have direct conversations with the people who've leveled up, make AI enablement available to everyone fairly, and update job expectations going forward rather than trying to retrofit them onto existing roles.
If you're thinking about how AI adoption should factor into your team's structure, roles, and compensation — and you want a concrete framework for your specific situation rather than general advice — let's talk. That's exactly the kind of conversation we have with owners who are ahead of the curve on this.