When Gen AI Doesn’t Deliver: Why Frustration is Part of the Process

We often hear about the breakthroughs where Gen AI saves time, unlocks insight, and helps teams work smarter. I’ve often been that person on my soapbox, waxing lyrical about what I’ve been able to achieve with it. But what happens when it doesn’t go so swimmingly?

Recently, I found myself deep in an Excel challenge, using Gen AI to build and debug something in PowerQuery. As with most of my endeavours, it started with a great deal of promise. I came armed with clear intentions, helpful prompts, and the confidence of someone who works with Gen AI daily. But gradually, I began running into snag after snag, and the route to success became muddy. I rephrased, reframed, and chain-prompted in a valiant effort not to concede. Eventually, the only thing that stopped me was running out of chat interactions altogether.

Rob: 0, GenAi: 0.

More frustrating than the tool itself was the sense of time wasted, and the creeping thought that I should’ve just solved it myself or asked a colleague.

The Reality Behind the Gen AI Promise

Gen AI is reshaping how we work. But amidst the excitement, it’s important to acknowledge a more grounded truth:

Gen AI isn’t infallible. It’s not always accurate. And it’s certainly not always efficient - especially when time is tight.

Having lived and breathed fast-paced environments, the expectation has always been to deliver effectively at speed. Yet even the most advanced tools can’t deliver without direction, clarity, and importantly, patience.

That’s not a failure of the technology. It’s a reminder that how we use Gen AI matters just as much as what we use it for. That’s why I wanted to share some reflections and the mindset shifts I’ve adopted when facing challenges with the tech myself.

Three Practical Mindset Shifts for Professionals Using Gen AI

If you’ve ever found yourself in the same boat of being frustrated with a Gen AI assistant, you’re far from alone. Here are three of my takeaways to help you make the most of these tools (without hurling your device in the process):

1. Start with clarity, not just curiosity

Gen AI performs best when it’s grounded in specifics. A vague prompt will lead to a vague response. A well-framed question or challenge that's complete with context, constraints, and a clear goal can save you multiple prompts and iterations. Think of it as briefing a team member: if you wouldn’t ask a colleague to “just fix this” without context, don’t ask Gen AI to.

2. Take a pause when progress stalls

When the answers feel off or inconsistent, resist the urge to keep pushing. (Learn from my mistake here.) Instead, step away. Break the task down as if you were explaining it to a peer. Often, the clarity you need doesn’t come from more prompts, it comes from rethinking the problem.

3. Know when Gen AI isn’t the right tool

Not every task is Gen AI-ready. Some problems are better solved through search, experience, or a five-minute chat with someone who has been there. This isn’t a weakness, it’s a strength. It’s what we did long before this technology became our go-to coach. The best professionals know when to change tack.

Moving Forward: Human Insight Still Leads the Way

The rise of Gen AI isn’t about replacing thought, it’s about enhancing it. And sharpening that skill will be essential as Gen AI becomes more embedded in the way we work. There will be moments when this technology accelerates your output, and moments when it slows you down. What matters is your ability to navigate both with intent and knowing when it’s time to adjust your approach.

So if your next Gen AI session doesn’t yield results on the first try, don’t be discouraged. Invest a little more up front. Step away when needed. And remember: sometimes the best guidance still comes from human experience.

Efficiency isn’t just about tools, it's about judgment.

And judgement, no matter how advanced AI becomes, is still very much a human skill.

---

Credits

Author: Rob Anderton

Editorial: Rob Anderton / OpenAI. (2025). ChatGPT 4o [Large language model]. https://chat.openai.com/chat

Images: OpenAI. (2025). ChatGPT 4o [Large language model]. https://chat.openai.com/chat