AI / Technology / Thinking / Work
AI can produce something remarkably quickly. That does not mean the thing needed to exist.
The more I use AI in actual work, the less interested I become in treating prompting as the main skill. Prompting matters, obviously. Clear instructions usually produce better results than vague ones. But the more capable the tool becomes, the more obvious it is that the difficult part often happened before the prompt.
What problem are we solving? Which information can be trusted? What is the constraint? What does good look like here, specifically? What should the system remember? Who checks the result? What are we willing to automate, and what still requires judgement?
Without those decisions, AI does not remove ambiguity. It reproduces it faster.
A system can generate fifty campaign ideas in the time it once took to make five. That sounds like progress until nobody has decided what perception needs to change, what the audience already believes, or which idea would actually be useful to the brand. More options do not automatically create more direction. Sometimes they simply create a larger pile of things nobody has a reason to choose between.
This is why I find AI most useful when it sits inside a process rather than pretending to replace one. I use it to research, compare, challenge, organise, iterate, document, and sometimes produce. It can widen the field quickly. It can reveal patterns I might have taken longer to notice. It can make a rough direction visible early enough to argue with it before too much time is spent polishing the wrong thing.
But I still need to decide whether the answer makes sense. That part is easy to underestimate because judgement does not look as spectacular as generation. There is no dramatic reveal when somebody quietly notices that the premise is wrong, the source is unreliable, the tone is off, or the beautifully produced answer solves a problem nobody actually had.
As production becomes easier, judgement becomes more visible by contrast. The interesting question for me is no longer only what AI can make. It is what becomes more important once making is no longer the scarce part.
AI expands capacity. It does not remove the need to know what you are doing with it.