AI is no longer a buzzword—it’s a browser tab.

And for those of us in reward, that’s both exciting and a little daunting.  The good news? AI can help us with our reward work; from benchmarking and job descriptions to policy drafting and market scans; tasks that once took hours can now be done in minutes—with the right tools.  Need to compare salary data across sources? AI can spot patterns. Drafting a pay policy? It can give you a first pass. Want a market snapshot? AI can summarise the latest reports while you grab a coffee.  Used well, AI isn’t just fast—it’s insightful. It can spot trends, challenge assumptions, and offer fresh perspectives. Think of it as the intern who never sleeps—the extra pair of hands you wish you had during budget season.

But… Garbage In, Garbage Out

AI is only as good as the data you feed it. If the sources are outdated or inconsistent, the outputs will be too.  This means you need to be careful what you ask it for.  Before trusting an AI-generated insight, ask:

• Is the data credible and anonymised?
• Is it relevant to your sector and structure?
• Are you benchmarking against last month—or last year?

AI won’t flag these issues. That’s your job. Read it, check it, test it.

Beware the Illusion of Precision

AI sounds confident—even when it’s confidently wrong. I’ve seen slick salary benchmarks and job matches that fall apart on closer inspection of the data. Why? Because AI is trained to sound authoritative, not to admit uncertainty. It wants to be helpful, so it will find and use out of date data or make assumptions based on what it can find. 

So challenge the outputs. Ask where they came from. Check the assumptions. Make sure they are fit within acceptable parameters.

Be the Human in the Loop

AI can crunch numbers, but it can’t read the room. It doesn’t know your culture, budget, or how a policy will land with your exec team.

That’s where you come in.  Reward decisions aren’t just technical—they’re relational. AI can help you get to the options faster, but it can’t choose the right one. Don't blindly follow where it leads you.  Interrogate it, sense check the responses, impose your expectations

Don't Forget Compliance Over Code

AI in reward isn’t just a tech issue—it’s a compliance one. Before using AI to benchmark pay or analyse equity, ask:

• Is the model trained on diverse data? (so many issues around developed bias in AI)
• Can you explain how it reached its conclusions?
• Are you sharing personal data with third-party tools?

Regulators are watching. So are your employees. Do your due diligence. Document your decisions. Keep human accountability front and centre.

Prompt IT Check IT Remind IT

Tell it everything you want, explain in detail, right down to the style and nature of the output.   Give it access to your public persona if you want to it write in your/the company voice.  Check the output and identify any issues.  Where it has missed an action tell it to go back and do it.  You cannot hurt its feelings, be blunt/clear/precise. 

Your AI Reward Readiness Checklist

Thinking of using AI in your reward work? Start here:

• Do you know what tasks you want AI to support? (Start with an easy one and build from there)
• Do you have comprehensive prompts?  (We have a suite of these)
• Have you vetted the data sources? (Prompt it to name and date them)
• Can you explain the tool’s logic?
• Do the outputs align with your strategy?
• Have you considered legal and ethical risks?
• Is your team trained to challenge the results?

If you answered “no” to any of these, don’t panic. Start small. Build confidence. AI is here to help—not to take over.

Still wary - I'm happy to talk it through.  I have some more in-depth guidance - I will even share my favourite prompts.