There’s a specific kind of AI disappointment that most professionals have experienced at least once. You type something into ChatGPT, get back a response that’s technically fine but somehow completely useless, shrug, close the tab, and go back to doing the thing yourself. You file AI under “overhyped” and move on.
The problem wasn’t the tool. It was the approach.
Most people use AI like a vending machine. Insert request, receive output, accept or reject. If the output is mediocre, they assume that’s just what AI is. What they’re actually experiencing is the result of an undercooked prompt, and the gap between a bad prompt and a good one is enormous.
The required perspective change is this; start treating AI like a thinking partner rather than a search engine. Push back on responses that aren’t quite right, ask for alternatives, tell it the response is too formal, or too vague, or that it’s missed the point entirely. The quality of what you get back is almost always proportional to how specific and engaged you are as a collaborator.
The other shift that makes a huge difference is context. Before you ask AI to do anything, tell it who the output is for. “Summarise this report” produces something generic. “Summarise this report for a CFO who is sceptical of the budget increase and needs to be convinced in under 200 words” produces something you might actually use. Audience-awareness is the single fastest way to improve AI output quality.
There’s also an awkward middle phase that most people get stuck in. You’ve used AI enough to know it can be useful, but not enough that it feels instinctive. Prompting feels effortful. Results are inconsistent. It’s tempting to conclude the whole thing is overrated and quietly revert to old habits. Don’t stop there. This phase is normal and it passes. The prescription is more repetition on lower-stakes tasks until the habit becomes automatic.
One practical move that accelerates this; start saving the prompts that work. When a prompt produces something that genuinely impresses you, keep it. Over time, this becomes a personal library of templates you can adapt rather than starting from scratch every time. The professionals who have built a decent prompt library can reliably complete in fifteen minutes what used to take an hour, not because AI has improved, but because they’ve stopped wasting time figuring out how to ask.
The professionals pulling ahead aren’t the most technical people in the room. They’re the ones who stopped dabbling and started using AI daily, on real work, until it stopped feeling like an experiment and started feeling like just how they work.
That transition is available to everyone. It just requires actually making it.
Scott Anderson
Scott is a product engineer and serial builder with over two decades of experience creating SaaS products and web software. As co-founder of Floats.ai and Sirius.dev, he brings deep technical expertise to the intersection of product design and engineering – turning complex problems into practical, scalable solutions. Floats.ai are sponsors of Hire Note.
