Remember the first time you used ChatGPT? It was magical and alarming. A chatbot capable of original writing. Haven’t seen that before.
Some HR practitioners quickly leapt on the AI train. Many others are optimistic about it but don’t know where to start.
We recently measured AI sentiment in compensation. With 20 percent of respondents “totally onboard” and 51 percent “cautiously optimistic” about artificial intelligence, its adoption might come sooner than expected.
Comp professionals should strongly consider introducing AI into their workflows. Why? Well, for starters, your CEO thinks your a Luddite.
A recent Gartner survey of CEOs ranked CHROs dead last in AI savviness among C-suite execs.
With this executive attention, HR practitioners need more than just enthusiasm. They need a practical roadmap for AI adoption.
If you don’t use AI regularly, this starter pack is for you. Even if you use it daily, you’ll find something here. We’ll cover everything from the risks of ChatGPT to selecting the right AI solution.
Let’s dive in.
ChatGPT-ing jobs? Tread carefully.
Some HR folks are already using ChatGPT for routine tasks like writing job descriptions.
If you’re only looking at the lowest-hanging fruit, you might be fine with open AI systems.
But remember that job descriptions are the bedrock of your comp structure. While you may speed up this task with ChatGPT, accuracy is essential.
When you benchmark jobs against poorly written, incomplete, or misleading job descriptions, you’ll end up with a framework built on a crumbling foundation.
Unlike purpose-built AI tools with transparent inputs, free algorithms use the entire Internet as training material. And you might be aware: not everything you read online is fact.
What about using ChatGPT for more advanced tasks like market pricing?
Say you’re looking for the salary range for a graphic designer in Austin, TX. You query a chatbot, and it spits out a reply.
The problem?
ChatGPT’s responses will sound entirely plausible, even if they’re utterly wrong. Using it for market pricing can sabotage your comp strategy. Without proper validation you might widen internal inequities or price yourself out of talent markets.
Additionally, any information you enter in ChatGPT might be available for public consumption.
Fast Company and other media outlets recently ran a story that sent chills up our spines. OpenAI (the company behind ChatGPT) allowed users to share chats that were indexed by Google.
While users had to opt in, it is unclear whether they knew other search engines would index their conversations. Within hours, OpenAI disabled this feature, calling it an “experiment.”
We don’t know when OpenAI will run another “experiment” like this, but it should give us pause. As a rule, you should only publish information in ChatGPT or other AI tools you’d also post to social media.
If it doesn’t belong on LinkedIn, it doesn’t belong on ChatGPT.
This is where building out a proper AI toolkit comes in — not just for avoiding the risks of free tools, but also to ensure it meets a real need.
AI use cases
Many HR practitioners have already dipped their toes in AI. If you’ve used it to draft a job description, congratulations on being part of the revolution.
This may feel like a small step, but tasks like writing job descriptions are the whole point of artificial intelligence. AI should handle your routine busywork, so you can focus your energies elsewhere.
Let’s explore AI’s current use cases.
Job descriptions
Are you sick of us mentioning job descriptions? There’s a reason why we keep bringing it up.
It’s low risk. It’s easy. And we’ve yet to encounter a comp pro yearning to write job descriptions.
While ChatGPT and other LLMs (large language models) can produce job descriptions, we’ve already mentioned the risks. So what are your options?
It's best practice to use job description generators in comp management software. This also allows HR to track changes and collaborate on drafts. If you’re using a free AI tool for job descriptions, you won’t have any oversight when the hiring manager accidentally decides a role requires a master’s degree.
Another plus: certain tools allow you to manage job descriptions and price jobs — all on the same platform.
Job match suggestions
Making survey matches is a hassle. Thankfully, AI is especially good at identifying semantic matches.
Instead of dealing with changing job codes and making participation matches manually, you can rely on an algorithm to do the heavy lifting.
How does it work?
Al sorts through thousands of jobs through natural language processing (NLP), analyzing keywords and picking up contextual clues. With automated match suggestions, an algorithm identifies similarities between your incumbent roles and participation matches.
It takes comp experts six minutes to correct one survey mismatch manually — AI chops this time down to seconds.
Market pricing
Top survey providers boast about the size of their datasets. And while it’s true the best surveys have thousands of participants and millions of priced jobs, their datasets are still limited.
AI market data isn’t.
Algorithms can price any job. With a large enough dataset as training material, AI prices jobs according to the differentials you select, such as industry and location. The tech is already advanced enough to challenge traditional job pricing methods.
AI for job pricing
HR practitioners are increasingly turning to AI for market pricing, but many are making a critical mistake.
They’re using free tools like ChatGPT or Google’s AI snippets, which pull salary data from sources like Reddit. Yes, Reddit. Not exactly the gold standard for compensation data.
Purpose-built AI job pricing tools are a completely different story. The best use millions of data points from HR-reported salary sources to deliver precise job pricing where traditional market data falls short.
Say you’re a manufacturer trying to price the role of Digital Identity Manager. Traditional market data gives you a handful of matches, but they’re mostly in IT and telecom in more expensive zip codes. Matches by your industry and location are scarce.
Comp professionals typically triangulate data and come up with a rough range. But “close enough” frankly isn’t enough. Overpayments balloon budgets, while underpayment poses retention risks.
AI resolves these issues.
Instead of struggling to find the relevant data, AI first identifies alternative jobs. Does an Identity and Access Manager II match your job description? Then it delivers exact market pricing with the data cuts you select, such as location, industry, or company size.
When the right data isn’t available or doesn’t match, the algorithm adapts. Let’s look at some examples of how it solves the thorniest pricing challenges:
Market pricing in low/no data markets
You’re pricing a Nurse Practitioner role in Blackwater, Missouri (Population: 160).
The problem? Blackwater doesn’t have any nurse practitioners. You’re the first.
Traditional market data doesn’t provide answers. So, what are your options? Use the national average salary for nurse practitioners? Throw a dart at some salary ranges?
With AI tools, you can easily price this role. Algorithms can identify nurse practitioner positions in similar geographies and offer strong benchmarks — even in markets with no relevant data points.
Eliminating the need for matching
Comp professionals labor to find the closest match for their internal roles. They pore over survey workbooks to pluck out that 70 percent match between a Digital Identity Manager and an IAM II.
The perfect match is elusive for all but the most common jobs. Job matching for market pricing is about confidence, not perfection, which is precisely what AI delivers.
With AI market pricing, you’re not chasing matches. Instead, you’re handed the best match with minimal effort.
Overcoming data dominance
Mega Corp Inc. floods your market with thousands of jobs across its huge network in your area. But you’re a mid-sized company with a few hundred employees.
Large organizations can overwhelm your market. AI weighs this data accordingly, ensuring that big employers don’t skew pricing.
If you’re going to do AI market pricing, the next step is choosing the right vendor. But these tools aren’t created equally, and selecting the best partner is more than just comparing features.
It means understanding how the technology works and the data it relies on.
Questions to ask your vendor
Companies must proceed with caution, pushing vendors for complete transparency about how AI is currently deployed and their planned roadmap.
Before introducing an AI solution, ask the following:
What problem does it solve?
Do you lack market data? Are you lingering on Indeed to find salary ranges? If so, AI market pricing might make sense.
On the other hand, if making survey participation matches drains hours from your day, automated match suggestions might be the way to go.
AI must be more than flashy tech. It needs to solve actual problems.
What are the model’s training materials?
Yes, we keep saying it, but ChatGPT makes stuff up. Some comp vendors’ AI products rely on ChatGPT as the backbone of their “system.” This is a problem.
But isn’t it getting smarter every day? Yes and no. OpenAI is improving its model, but it’s also hallucinating more often. One theory is that the more reasoning the model attempts, the greater its chances of providing incorrect responses.
Long and short of it: make sure you ask vendors about their algorithm’s training materials.
How will you measure performance?
What’s an acceptable standard for accuracy? AI tools should cut down on your workload, not leave you redoing tasks.
If you’re using AI to make job match suggestions, how often are you accepting them? 50 percent of the time? 80 percent?
Decide on a reasonable standard for evaluating performance and make sure your solution delivers.
Does the model learn from its mistakes?
Without a learning loop, AI tools will keep making the same errors or producing irrelevant results.
The most valuable models are those that accept feedback and course correct when they miss the mark.
By asking these questions now, you’re not just selecting a tool — you’re choosing a partner. The decisions you make in vendor selection today will determine whether you’re leading the AI charge at your organization or scrambling to catch up later.
Early AI adopters will reap the rewards
AI isn’t coming in compensation — it’s here. Whether you’re testing the waters with job descriptions or ready to jump in with automated market pricing, the most important thing is to get started.
You can begin small. Some vendors (ahem) will even let you price a job using a verifiable AI algorithm for free. Your willingness to experiment now sets you up for later wins.
The question isn’t whether AI will change compensation. It will. It also isn’t whether your job will change. Also, yes. Instead, the real question is how you’ll prepare. Are you ready to advance your compensation program with artificial intelligence? Or will you wait until your manager comes knocking on your door asking why you’ve waited so long?
Want to know how to get started? Take a peek at our webinar: "How AI is reshaping pay decisions."
.png)
