Today’s HR professionals are already using next-gen AI in fun, creative, and exciting ways. For many, however, the real challenge comes in leveraging this technology to its fullest extent.
I recently sat down with Paul Carney, founder of Perpeta, and Mark Feffer, founder of WorkforceAI.news, to discuss modern AI systems, their effect on the HR industry, and how HR professionals can make the most of their AI tech.
Capturing the Value of Current- and Next-Gen AI Tech
“What we're doing with AI and how we're using it today is going to look extremely different in about two to three years.” – Paul Carney
Given the evolving nature of AI in general, some find it difficult to embrace this technology while it’s still in its infancy stages. Even if they’re interested in AI in the first place, many don’t want to invest the time and money into learning a new system that is only going to change in a few years.
However, being amongst the early adopters of powerful new technology – such as AI – certainly has its benefits.
“I remind people in HR: we don't just have to figure out how to use AI for ourselves. We've got the accountability and responsibility to figure out how the organization is going to use this – which means training, development, and policies.” – Paul Carney
Organizational leaders can begin taking advantage of current-gen AI while making preparations for future upgrades and innovations. As a company, your goals for AI integration will remain largely the same – even if you have to provide additional training or modify your policies as the technology improves.
“I think the people that do the best with AI are the people who start out knowing what they need to get from it.” – Mark Feffer
1. Building a Chatbot Replacement in Minutes
According to recent studies, 69% of companies already use chatbots and virtual assistants on a day-to-day basis – making them amongst the most widely used AI tools today. But did you know that you can use AI to create a customized chatbot for your organization?
In most cases, someone who is skilled with the current generation of AI software can have a chatbot up and running in approximately 60 minutes.
“There’s a use case here to effectively put a chatbot on top of your employee handbook or similar documentation. Now, what’s actually pretty cool is you can use one of these extremely cheap and easy AI tools to basically learn your documentation and answer these questions. You can build your own bot effectively in an hour.” – Phil Strazzulla
ChatGPT can also be used to create a virtual assistant (VA) for you or your organization. Like other virtual assistants, the AI-driven VA is capable of responding to inquiries, transcribing notes from meetings, creating blog posts, and much more.
But chatbots and VAs are amongst the more advanced applications of modern AI technology. In order to make the most of these systems, the user really needs to have a basic understanding of how current-gen AI systems work, how they analyze patterns and trends, and how they respond to inquiries. Moreover, they need to understand the concept of AI prompt engineering and how they can use this strategy to their advantage.
2. Write and Post New Job Openings to LinkedIn
The prompt engineering process involves the use of specific and intentional instructions, referred to as prompts, in order to help guide the responses of a generative AI model – including large language models (LLMs) like ChatGPT and others. When given specific instructions, most AI systems tend to produce better, more targeted results. Prompt engineering is often used to help organizations (and HR professionals) write and post new job openings to LinkedIn or other online job boards. In the presentation, Paul Carney offers up a prompt to create two separate LinkedIn posts.
- For the first post, he instructs the AI to target job seekers who are returning to the workforce.
- For the second post, Paul instructs the AI to create a second post targeting an audience of established HR professionals who are looking for a more challenging environment.
He also tells the AI to include relevant emojis and hashtags for each, which creates a more eye-catching post while also maximizing his reach across the LinkedIn network.The result is two posts that are essentially the same – only targeting two different types of job seekers. Per LinkedIn algorithms, the first post will appear in the feeds of returning HR professionals while the second will appear for (and appeal to) those who are currently working in the HR industry.
“By separating and taking one job posting and creating numerous types of targeted audiences, the words, the phrases, and the things you’re saying here are going to cause LinkedIn to put it in their feed – which is one of the advantages you get, the power you get, by doing something like this.” – Paul Carney
3. Create a Job Interview Coach for Your Team
In this use case, Paul uses another specific AI prompt: “You’re an interview coach helping us prepare for interviews by simulating responses and generating follow-up questions.”
He then provides a follow-up prompt: “Review the job description and ask me what my first question is. When I ask the question, you simulate the interviewee’s response based on the job requirements. After you provide the response, provide three possible follow-up questions to allow me to explore.”
Finally, Paul uses a third prompt: “We will conclude the simulation as you provide answers. We will conclude the simulation when I say ‘thank you for your time.’ After the simulation concludes, you’ll provide me with suggestions for improving my interview techniques.”
Not only do these prompts tell the AI exactly what to do, but they also tell it where to find the relevant information.
Prompts like this also give the AI some freedom in coming up with three follow-up questions based on the simulated interviewee’s responses. The resulting job interview coach can be reused, as many times as needed, for onboarding, training, or even just for fun.
Prompt Engineering Best Practices for HR Pros
“It always comes down to the prompt because no matter what these tools look like in the future, how you interact with them and how to deal with these systems is the key. Practicing now is great because eventually, when those tools mature enough, you’re going to be way ahead of the game.” – Paul Carney
Paul wraps up his crash course on AI prompt engineering by laying out a few basic best practices.
- Remove all ambiguity from the prompt and be as concise as possible
- Create reusable templates that can easily be copied and pasted, or modified, as needed
- Try out different AI systems to find the one that produces the best results for you and your team
These best practices provide a great start to AI prompt engineering. However, they’re really only the beginning of your path to achieving full AI integration.
