Would you believe that Elvis Presley predicted the current state of Generative AI back in 1968? Crazy, right? But the King of Rock 'n' Roll might as well have been singing about today's AI technologies when he sang the line “A little less conversation, a little more action please.” At this moment, this line sums up perfectly the current state  of Large Language Models (LLMs) like ChatGPT, Gemini, and Claude.

Given the right prompts, these AI models can assemble responses that are not only helpful but can even border on genuinely insightful. They can draft emails, create content, summarize reports, and even generate coding scripts—all of which seem like actions, right? But are they really? While these AI models generate text-based outputs (the 'conversation'), the 'action' would involve these models taking steps that directly execute tasks without needing human intervention—something they are currently not capable of (yet).

Integrating AI into roles like Human Resources requires careful consideration, especially given the crucial need to interpret nuanced information accurately.

While Generative AI can assist in tasks such as resume screening, it is important that we remain in the loop to oversee these processes. This ensures that subtle but critical details influencing candidate evaluations are not overlooked, preserving the integrity and effectiveness of hiring decisions.

Ultimately treating AI, like a well-meaning intern, is an effective way to ensure the appropriate balance in your use of it. With that in mind, what tasks can be simplified or streamlined using Large Language Models in their current state.

  • Synthesizing Compliance Language for Job Descriptions

  • Creating Data-driven Recruitment Strategies 

  • Analyzing Employee Feedback

These tasks are just a fraction of what Generative AI can do. Yet these tasks, and all others, are reliant on effective prompt building in order to return useful data. And that is where the current state of LLM’s falls short. The ‘action’ part is likely to be covered soon with the introduction of Large Action Models. However, that leaves the prompt as the potential kryptonite that inhibits wide adoption of these tools in a business setting.

We have all experienced this in a current or past work setting where someone claims this magic tool will make your life better. And while it may have turned out to be true, there is a learning curve. And if you are already traveling at Mach 10 with your hair on fire on a daily basis, finding time to learn how to write that magic prompt may not be your highest priority.

You should think of a prompt as if you were getting ready to rub Aladdin's Lamp. What do you wish for? Think carefully as we all know how some of those wishes end up.

For example, here is the anatomy of an effective prompt for your AI intern to build a job description:

1. Have a specific objective

  • "Create a detailed job description for a Project Manager focusing on digital transformation projects, emphasizing responsibilities and required skills."

2. Give detail 

  • "Include responsibilities such as overseeing project execution from planning to closure, and skills like proficiency in digital tools, leadership abilities, and experience managing large teams. Make sure the text is gender-neutral and includes a statement on the company’s commitment to diversity and inclusion."

3. Give context

  • "This job is for a mid-sized tech company known for its collaborative culture and innovative approach to technology solutions. The project manager will work closely with software development and marketing teams."

4. Give it Style

  • "Use a professional yet approachable tone that reflects our company’s supportive and forward-thinking workplace environment."

5. Don’t forget about the Compliance and Legal Considerations

  • "Ensure the description complies with the Pay Transparency Act by clearly stating the salary range and affirming that the final offered salary will depend on experience and qualifications."

If you were now to cut and paste each of the above bullet points in quotes into ChatGPT, there is your prompt. But whew! Really? While this structured recipe can be customized to any task, is it worth the investment of time to learn how to craft a good prompt? It can be. Especially if you are an early adopter, enjoy learning, or are simply curious.

The hype of AI, both good and bad, is focused on the future. But there needs to be a productive present as well.

To democratize the use of LLMs in HR, it is essential to develop tools and interfaces that simplify prompt engineering. By reducing the complexity and time investment required to leverage AI, we can make these powerful capabilities accessible to all HR professionals, not just the early adopters or the technically adept.