Prompt Engineering: When to Ask a Human vs. AI

Over the past six months, as I’ve increasingly woven AI into my workflow, I’ve noticed a new habit forming: Every time I have a question, I stop and ask myself:

You’re probably laughing on the inside. What a silly question, you’re probably thinking. Of course humans are better. And in many cases, you’d be right. But not always.

If you think AI’s generative powers are scary or unsettling, just try asking 10 different humans for career advice. You’ll walk away with your head spinning from all the conflicting opinions, not to mention a packed calendar full of follow-up conversations. It's overwhelming.

That’s not to say humans don’t have a crucial role in your job search (or any process, for that matter), but when you just want the basics—no hot takes, no opinionated tangents, no extra fluff—sometimes AI is the more efficient option.

Since I use AI for almost every part of my workflow these days, I first turned to ChatGPT to help me come up with a recent example of when I went to AI first (instead of a human). It reminded me that I’d recently used AI to refine my short-form stories for job interviews, specifically using the STAR technique (Situation, Task, Action, and Response).

This was a perfect example of how AI fits into my workflow. While I could have sought advice from my network, AI was more than sufficient to help me tighten and improve the stories on my own.

Then I then asked ChatGPT to analyze how I typically use AI vs. humans. It was surprisingly accurate.

Then I worked with ChatGPT to come up with a side-by-side comparison of how I use AI vs. Human input in my project work. 

Through this process, I've become more aware of the strengths and weaknesses of both human and AI input. Each offers something unique, depending on what I need at the moment.

Let’s start with humans:

Now, onto the AI:

Interestingly, you’ll see the inclusion of bias on both lists. 

We often focus on the inherent bias in large language models, rooted in their training data, but we talk less about the overt biases humans bring to the table. Of course, human biases, shaped by our personal experiences and perspectives, can be even more direct and pronounced than the subtle biases in AI.

I saw the downsides of human bias play out clearly at a panel discussion a few months ago.

As one of six speakers, I was asked to share insights on the job outlook for fast-growing roles, specifically in the tech sector. The moderator asked us all the same question: "Can you share the industry trends you're seeing and what skills job seekers need to learn?"

The responses were strikingly different:

Human Prompt: "Can you share the industry trends you're seeing and what skills job seekers need to learn?"

Human Responses:

• Three panelists gave broad overviews, citing labor market data.

• Two focused on specific roles, diving deep but missing the bigger picture.

• One person spoke exclusively about their own organization.

From the audience's perspective—likely educators without deep industry knowledge—this must have been confusing. They needed clear, actionable information to pass on to their students or colleagues. But the range of responses made it hard to piece together a coherent picture.

I started to wonder: Was the problem with the question itself, or the way each of us interpreted it? If the goal was to provide a level playing field across industries, we as humans fell short. The responses were too varied—perhaps a result of overly broad prompts or personal biases.

It made me think: A custom AI agent could have provided consistent, clear answers about industry trends, leaving the humans to add the deeper insights and details that only come from lived experience.

Ultimately, it’s not about choosing between humans and AI—it’s about knowing when to rely on each for the best results. 

In my own workflows, I’ve found that AI is great for efficiency, quick feedback, and objective answers, especially when I want to bypass bias or conflicting opinions, or quickly grasp a general overview of a new topic. Given the wide-ranging nature of my fractional work, even these micro-efficiencies add up, saving me hours throughout the day.

But then there are times when you just need a hug. Or the emotional insight that only humans can provide. When I need nuance, provocative “hot takes,” or an energizing brainstorming session, that’s when I get back the IRL thing.

After all, there’s still no match for what a few over-caffeinated people can accomplish with a whiteboard, some inspiration, and a couple of pizzas.