If You Are Using or Deploying AI Tools, You Better Be Aware of Tokens

Many HR professionals that are using artificial intelligence (AI) tools for their tasks and projects are unaware of the technology’s hidden costs, which are tied to tokens.

Tokens are the building blocks of what’s inputted into an AI tool (from a 10-word question to a 200-page document) and what’s generated in response (e.g., a three-sentence statement, a meeting summary, a slideshow, onboarding materials). While token measures vary across AI models and providers, a commonly used average is one token represents about four written characters. And, they carry a tangible cost that companies and their functional areas may not fully grasp until they quickly hit a predetermined limit — or get an overage bill from their AI vendor.

Creating a performance review template? That takes tokens. Asking a chatbot to read a half-dozen health plan documents to determine which one best fits each team member? More tokens. Uploading hundreds of job descriptions and running a comparison scenario of two types of job architectures — and then regenerating the full recommendation report after the AI tool suggests a nominal change? Tokens on tokens on tokens.

Access bonus Workspan Daily Plus+ articles on this subject:

‘Trouble With Tokens’

You can liken these data units to metered electricity usage, or cellphone minutes back in the day, said Rob May, the CEO of Neurometric, a token engineering platform that helps companies optimize AI usage. He’s also the author of “The Tokenminning Manifesto,” a guide that flips the script on the recent “tokenmaxxing” movement that linked prolific token usage to AI innovation.

“HR people can really get into trouble with tokens, because HR is a very document-heavy part of the organization,” May said. “What seems like a very simple question that’s passed along with 70 pages of documents can suddenly really get expensive.”

How big of a financial hit are we talking about? For context, consider three recent examples:

JPMorgan reported that some investment analysts were running personal token budgets into the tens of thousands of dollars, a handful past $100,000.
One Uber executive burned through the company’s entire 2026 AI budget in four months.
A four-person startup posted a $113,000 monthly bill from a single model provider.

(Also, check out this article from consulting firm Alvarez & Marsal on updated token pricing.)

Quick Tips for HR Pros

In considering the financial impact of tokens, keep the following guidelines in mind, said Sara Hillenmeyer, the vice president of AI and data science at compensation software and data company Payscale:

Your conversation history with an AI tool is reused with each fresh query. If you don’t need the context, start a new prompt. Share files and results in collaborative AI workspaces or projects to avoid repeated inputs.
Consider whether your task can be completed using a cheaper model — and build an internal AI system that allows you to swap between models.
Tokens tied to AI-generated output are usually significantly more expensive than those measuring user input.

“It’s cheaper to give more context and ask for more information up front than it is to, for instance, repeatedly generate the same output,” Hillenmeyer said.

May also offered the following advice:

Just because you can include massive files with an AI query doesn’t mean you should. If you know your question pertains to Section 4 of the employee handbook, upload only those pages rather than the full document.
Explicitly ask AI vendors to outline token costs and billing structures.

Employees’ current awareness of tokens may be limited to “tokenmaxxing” — and that’s if they’re familiar with the concept at all, said Julia Dhar, a managing director and partner at Boston Consulting Group and the co-author of the book “How Change Really Works: Seven Science-Based Principles for Transforming Your Organization.”

“Before we had this very noisy, public conversation about token usage over the last several months, it would be reasonable to say that most people in most companies had no sense of token economics,” Dhar said. “If you’d asked, ‘How much does a token cost? Do you know that we pay by the token?’ that knowledge was not widely available inside companies. We can’t fault anyone for that.”

But that level of visibility and knowledge is needed — by you, the HR function and the organization.

Want to Learn More About AI? Examine These WorldatWork Webinars

Click here to view the full slate of webinars.

Editor’s Note: Additional Content

For more information and resources related to this article, see the pages below, which offer quick access to all WorldatWork content on these topics:

If You Are Using or Deploying AI Tools, You Better Be Aware of Tokens

‘Trouble With Tokens’

Quick Tips for HR Pros

#1 Total Rewards & Comp Newsletter