A conversation at work recently made me stop and think.
Someone asked, "When AI gives you a poor answer, do you start a new chat or do you rewrite the prompt?"
All of us in the room had an opinion. One colleague was so convinced he was right that he even said, "Bet?" I laughed and said, "Happy to."
Because I was certain I had the answer.
But the conversation got me thinking. The more I reflected on it, the more I realised we were debating the wrong problem.
The longer I've worked with AI, the more I've realised that the quality of the output rarely depends on whether you started a fresh chat or rewrote the prompt.
It depends on whether you understood the problem well enough in the first place.
I've seen exceptionally well-written AI outputs that completely missed the point. They looked impressive. The language was polished. The structure was flawless. But they didn't account for the context, the stakeholder dynamics, the political sensitivities, or the history that never exists inside a prompt.
I've also seen every kind of AI user over the past couple of years. Some speak about it with tremendous confidence. Some quietly experiment behind the scenes. Some have spent years exploring the technology. Others have only recently started. Experience with the tool, by itself, doesn't necessarily translate into better outcomes.
I've also seen people who wouldn't call themselves "AI experts" get exactly what they needed because they knew what information mattered, what assumptions to challenge, and what still needed to be validated.
That difference has very little to do with prompting.
It has everything to do with judgement.
Perhaps that's why I'm not overly worried when people ask whether AI will replace experienced professionals.
The value of experience has never been in writing the first draft. It's in knowing what deserves attention, what can be ignored, what needs another conversation, and what simply shouldn't be delegated to a machine.
Technology will continue to evolve. The interfaces will become simpler.
Prompting itself may eventually become invisible.
Judgement, however, is accumulated far more slowly.
Over the past year, I've been fortunate to learn from people who have been incredibly generous with their knowledge.
Andrew and
Shahnaz challenged the way I think about AI and encouraged me to experiment beyond my comfort zone.
Tanushvi has been an invaluable sounding board for countless brainstorming sessions, reminding me that some of the best ideas still come from good conversations. And my time at
Archetype exposed me to
Marker Collective and AI-first ways of working that fundamentally changed how I approach the technology.
Building custom GPTs despite having no coding background taught me one thing above all else: expertise isn't about having all the answers.
The tools will keep changing.
Curiosity, humility, and sound judgement never go out of style.
#CareerJourney #Leadership #PeopleFirst