Is ChatGPT The Must-Have AI Tool for HR Companies?
ChatGPT. Millions have embraced it. Many are still skeptical. But it’s here, it’s not going away, and the technology has clear benefits for the HR and TA space — when used ethically and leveraged to produce optimal outcomes.
Cliff Jurkiewicz, VP of Strategy at Phenom, shares how the smart, appropriate use of ChatGPT can drive efficiency and productivity so HR team members can focus on the people part of their jobs.
In fact, we used ChatGPT to write the abstract for this episode, which we’ll get to a little later!
Read on for insights from Jurkiewicz — a 30-year veteran of the AI field — who shares specific use cases, optimization techniques, and ethical considerations around ChatGPT. You can also check out the full episode of Talent Experience Live right here.
Why is ChatGPT so successful?
Released to the public in November 2022, ChatGPT captured over one million users within five days.
For anyone who needs a quick recap on ChatGPT: it’s an AI-driven natural language processing tool — or chatbot — that can be trained on individual writing styles and used to assist in writing content such as emails, social media posts, job descriptions, and, ahem, college essays, among many other things. This technology can also assist with research and analysis.
“When people see a tool they think can work very well… and they can personalize where it fits into their workflow, I think you’ve got a winning combination,” Jurkiewicz noted.
ChatGPT is built on a technology known as generative artificial intelligence — or GAI for short — which are essential tools that generate content using AI.
AGI, on the other hand, stands for artificial general intelligence, referring to AI that has cognitive abilities that imitate human thinking (think Agent Smith from The Matrix or C3P0 from Star Wars). AGI is not widely used today.
So, who is the ideal user for ChatGPT? Any industry that needs to analyze problems or generate any sort of writing can benefit from ChatGPT, Jurkiewicz said. “I can’t think of a vertical where it can’t be applied,” he emphasized.
Related reading: 6 Companies Successfully Using AI in Their Recruiting Strategies
What are some current drawbacks?
Despite the tool’s vast potential, like any form of technology, it’s not a magic bullet. Currently, GAI’s ability to respond correctly to any given prompt is about 50% to 70%.
A big reason for that? You might be surprised to hear that ChatGPT is not connected to the internet (although it’s available online). Rather, it was trained on data from 2021 to 2022.
“GPT” stands for generative pre-trained transformer. Again, ChatGPT has been pre-trained on a subset of a year’s worth of the world’s data. The tool’s output is based on a large language model (LLM). Given any sequence or length of words, an LLM is an algorithm programmed to predict and then generate text and other content based on that sequence.
“[LLM] is not new… what’s new is that it’s learned the writing styles to a degree of precision and accuracy that it can actually be used in the everyday space,” Jurkiewicz said.
The key takeaways here? Know that ChatGPT isn’t 100% accurate, and that using it to its full potential takes some art and science.
What are some tips for optimizing ChatGPT?
Leveraging a GAI tool like ChatGPT involves more than giving a simple command (e.g., “Write a job description for a customer service representative”) and then copying and pasting.
You’ll get the best results by feeding the bot more specific information and instructions, and then sifting through responses and editing the text for a polished (and accurate) final product.
When we asked the tool to write the abstract for this episode, for example, our first instruction was: “Write an abstract for a livestream episode featuring a guest from an HR tech company who talks about the benefits of ChatGPT.”
The tool created an almost 300-word abstract, but it was missing a few details like Jurkiewicz’s name, the title Talent Experience Live, and why the tool is important for HR companies to use. In order to get it more exact, we had to rephrase our request to include a more limited word count and the specific names that we wanted included.
“We’re moving from content creation to content curation,” Jurkiewicz said. “You’re going to curate the response, and it might take several different takes to get it right.”
Here are some key phrases to curate more accurate responses when interacting with ChatGPT, according to Jurkiewicz:
“Act as if” / “Act like.” Here, you’re giving significant context to the tool. For example, you might say, “Act as if I’m a hiring manager in a distribution center with 20 years of experience and create interview questions for a frontline distribution center employee who’s going to drive a forklift.”
“Analyze [x].” With this phrase, you can prompt ChatGPT to refine, for example, a job description: “Analyze my job description for a forklift driver and tell me if it’s inclusive. Do I have the right keywords? Am I missing anything?” Then you can tell the tool to rewrite the description based on its analysis.
“Provide [x].” This prompt gives it a specific task: “Provide an analytical breakdown of the top performing career sites over the past year.”
“Based on x, give me y.” This prompt is helpful because you’re feeding it data: “Based on last year’s job opening rate, predict this year’s percentage.”
“Using this model / Using this as an example, write me [xyz].” This can be helpful when you need to write or follow specific codes or models. For example: “Using Python Programming Language, write me [x,y,z].”
Jurkiewicz also recommends giving it 5-10 samples of work you’ve already written, and asking it to analyze and remember that style for future reference. That way, you won’t need to spend as much time editing and revising. You can also ask ChatGPT to use all five prompts in one request, leaving you with an answer that should be fairly accurate.
Should job candidates disclose usage of GAI tools?
At what point does assistance from a GAI tool such as ChatGPT become a misrepresentation of a user’s true abilities?
Consider how Photoshop changed photography. Or how Auto-Tune revolutionized music production. Even with this new technology, people still need to start with their own talent and their own knowledge, Jurkiewicz pointed out. “You still have to have ideas. You still have to know how to use the tool.”
Is it necessary to disclose how much of what you’ve created was generated by the tool versus your vision? “At the end of the day, I think it plays itself out in the work product itself,” he continued.
So, the big question for hiring organizations: How do you handle the probability that job candidates will increasingly use ChatGPT to generate cover letters?
The way Jurkiewicz sees it, hiring managers should look at it as an opportunity to assess a candidate’s aptitude and creativity with technology. “I would ask them to use ChatGPT to create their cover letter and see how far they can take the tool … [to see if] they’re using this innovative technology in a really smart way.”
Related reading: AI and Amazing People: An Inseparable Duo
What are some examples of HR use cases for ChatGPT?
Generating interview questions. ChatGPT can create in-depth interview questions based on job requirements, a candidate’s profile, experience, and other relevant information. “[Questions that] allow me, as the candidate, to get very personal and show my value to the organization,” Jurkiewicz said.
Post-interview, the hiring team could feed the interviewee’s responses to the tool, guiding it to identify themes to highlight for a more personal, productive follow up conversation.
Writing job descriptions. This is where that “act like” prompt comes into play, giving the tool specific context to work with. For example: “Act like I’m a hiring manager with 10 years of experience with Python and I need to hire someone with 2 years of experience with Python.”
You can give word count limits, too, which is critical to writing job descriptions – and can be difficult to stick to. (No job description should exceed 750 words, Jurkiewicz pointed out. Ever.)
With these and other use cases, fine-tuning is key when it comes to using GAI tools. The user creates the “initial bulk work,” and then leverages the tool to curate the idea along until it’s generated an optimal final product.
“To me, that’s the most important thing — that we’re using that tool in the right way that’s going to give good results.”
How can organizations ensure ethical usage of GAI tools?
AI-driven tools like ChatGPT are a must for organizations that want to innovate — but ethical usage needs to be woven into the culture. “Create an environment where ethics can be discussed. Create the technology where ethics are the paramount principle being followed,” Jurkiewicz said.
Four key concepts guide ethical usage of GAI tools:
Is it ethical? Ensuring the tool won’t create bias in the organization is a top ethical concern.
Is it explainable? Be transparent about how and why you’re using ChatGPT.
Is it defensible? Organizations must be able to show that any use of AI is not causing discrimination. (If you’ve followed the first two principles, defensibility is much easier to uphold.)
Is it configurable? The technology must be configured so that it’s adding value to the organization.
The requisite AI question: Will ChatGPT render human input obsolete?
As with other AI-driven tools, the answer is “Not a chance,” according to Jurkiewicz. Human judgment and experience will always be essential.
“We want to innovate in really smart ways that include what we call the “human in the loop”, making sure that we are benefiting humanity in a way that allows us to progress.”
To illustrate, he posed another use case:
ChatGPT can be applied to help analyze pain points for a client, company, or market. When guided with the right human input, the tool can pinpoint challenges in seconds that humans might spend hours researching to uncover.
And those hours saved? They can be spent focusing on solving problems rather than looking for them.
Jurkiewicz’s final thoughts on ChatGPT? Embrace these technologies, but use guardrails, especially in the beginning:
Guide teams on use cases for GAI tools
Monitor how employees are using these technologies
Work with legal and compliance teams to ensure ethical usage
Be open and explain to people how the organization is using these tools
“When you do that, you’re creating space for a different conversation… a more focused, more important conversation.”
Get notified for all upcoming Talent Experience Live episodes here.
Maggie is a writer at Phenom, bringing you information on all things talent experience. In addition to writing, she enjoys traveling, painting, cooking, and spending time with her family and friends.
Get the latest talent experience insights delivered to your inbox.
Sign up to the Phenom email list for weekly updates!
