Using ChatGPT for job evaluation processes

As interest in AI grows, many Reward professionals are exploring how tools like ChatGPT can support compensation processes, particularly job evaluation and the implementation of Pay Equity regulations. So how can AI assist in this area, and what should be considered before implementation?

Job evaluation is the process used by organisations to assess the relative value of jobs to one another and determine appropriate pay ranges. Common methods include ranking, classification, point factor, factor comparison, and market pricing. Each method has its advantages but also inherent problems, such as subjectivity, rigidity, complexity, and potential for bias. For example, the ranking method is simple but highly subjective and unlikely to pass any gender-neutrality test, while the point-based method is detailed but complex and time-consuming.

As a result, there’s a lot to consider when using AI to support compensation processes and decisions.

Before you begin, it's essential to consider data security and privacy:

1. Does your organisation have ChatGPT Enterprise to ensure top-tier security, privacy, and advanced capabilities?

2. Can you confirm that employees cannot access sensitive data through a keyword search?

3. Are employees using personal AI accounts? Without enterprise access, some employees may resort to using personal AI tools for work, this can compromise data security and introduce unwanted bias. A recent HR Magazine article? [Link:?in comments]

]noted that one-third of employees use generative AI tools they personally pay for, highlighting the potential risks.

Consider reviewing your data access protocols and ensuring proper controls are in place before you consider using AI.

Implementing AI in Job Evaluation

The starting point for using AI in job evaluation is to apply or build a gender-neutral methodology that covers all relevant job functions. For example, one of my client’s has a manufacturing, sales, distribution, retail and wholesale business. It’s pretty diverse so building a methodology for them would be complex.

Next, define the roles in scope, gather job descriptions, and document the internal structure—just as you would in a traditional evaluation process. Once your evaluation criteria are set, you can craft prompts for ChatGPT to grade the job. For example:

Prompt: "?????? ?????? ?????????????? ???? ?????? ???????????????????? ?????? ???? ??????????????????????????, ?????????? ???? ?????? ???????????????? ?????? ?????????????????????? ?????? ?????? ?????????????? ??????????????????, ?????????????? ?????? ????????????????????????????????, ????????????, ?????? ???????????????????????????? ???????????????? ?????? ???????? ????????. ???????????????? ?????? ???????????? ???? ?????????????? ??????????, ?????????????????? ?????? ???????? ???????? ???????????????? ?????????????? ???? ?????? ?????? ?????????? ???? ??????????????. ?????????????????? ?????? ?????????????????????? ?????????? ?????? ???????? ????????, ?????? ?????????????? ?? ???????????????? ?????????????????? ???????? ?????????????????? ???????? ????????????????, ?????????????????????? ?????? ?????????????? ???????? ????, ????????????????????, ?????????? ???? ????????????????????????????, ?????? ???????????? ???? ?????? ????????????????????????. ???? ?????????? ?? ?????????????? ??????????????, ?????????????? ?????? ???????????????????? ???????????? ?????? ???????? ???????????? ??????????????"

This prompt encourages a more comprehensive evaluation, prompting ChatGPT to analyse specific job factors in relation to the overall structure and provides a clear justification for the grade.

Challenges and Considerations

Be prepared for multiple iterations before generative AI applications such as ChatGPT delivers a fair and meaningful response. Even then, line manager input is crucial to validate the AI's output and ensure it aligns with organisational expectations.

It is also important to question whether an AI-DIY (do-it-yourself) approach is worth the risks. ChatGPT itself warns that it can provide incorrect or nonsensical information that appears plausible but is not based on real data or facts (known as Hallucinations), so you'll need expertise to assess and trust the results.?

???????? ???????????? ?????? ????????????????: ?????? ?????? ?????????? ???????? ???????? ?????????????

Consider the value of professional support. There are many providers specialising in job evaluation who can offer reliable and cost-effective solutions. While AI can assist in grading jobs, it requires careful implementation, patience, and a robust methodology to be effective.

Using AI in job evaluation is possible, but it comes with challenges that must be carefully managed. Ensure you have the right tools, expertise, and safeguards in place to make the most of AI’s potential while protecting your organisation’s data and ensuring accuracy

I am indebted to Mike Doull for his support with this article. Mike is passionate about the convergence of data, AI, machine learning? and the future AI Economy and the most knowledgeable person I know on the subject.?

Photo credit: Victoria Gibbs