Do Not Automate the Average
The best use of AI is not to produce the average faster. It is to explore beyond the first obvious answer.
The easiest way to use AI is to automate the average. Ask for a landing page, get a landing page. Ask for a post, get a post. Ask for ideas, get the same ten ideas everyone gets.
That is useful sometimes. A competent first draft has value. But if we stop there, AI makes everything smoother and more generic at the same time.
The better use is to push the work past the obvious. Ask for extremes. Ask for constraints. Ask for contradictions. Ask for the version a cautious team would reject, then inspect why. Ask for five different mental models. Ask what would make the idea fail. Ask what is missing for a tired user, a skeptical buyer, a beginner, an expert, a person in a hurry.
AI is good at generating surface area. That means we can explore more territory before deciding where to stand.
Do not use AI only to produce faster. Use it to think wider before you choose.
This is especially important in design. The first solution is often the solution shaped by habit. AI can reinforce that habit if the prompt is lazy, or challenge it if the prompt asks for difference.
The point is not to be weird for the sake of being weird. The point is to make sure the final version was chosen, not merely accepted. Average output should be a checkpoint, not a destination.
If AI gives us speed, we should spend some of that speed on ambition. Try the stranger layout. Explore the clearer metaphor. Test the sharper headline. Build the better prototype. Let the tool remove friction from exploration, not standards from the final work.
— Victor