the appearance of a thing
You open the tool. You type a prompt. And thirty seconds later, like magic, a thing appears.
Maybe even two or three things.
Each thing appears to be different from the other things. And one of these things even interests you.
You feel pretty good about it.
And the thing that created the thing even told you that you did a great job.
You created a thing.
Or did you?
This magic thing maker gives the appearance of solving a problem. But it was unbothered by your prompt's incompleteness. It does not care if the answer is interesting, new, or actually useful. It just produces what statistically tends to follow prompts like yours.
What was produced was not crafted or creative. It confidently provided you with the appearance of creativity. The appearance of something both new and useful.
You created a thing.
Or did you?
As a designer, I have made mistakes and shipped them. Or shipped something I didn't know was a mistake and had to figure out why it was a mistake.
I have felt the deep seated discomfort of making a thing that looks right, but feels wrong. Where a detail felt "unresolved" due to unknown tension between a problem and the many solutions that could solve it. This accumulation of doubt and failure is not a side-effect of developing creativity. It IS creative development.
The thing maker has only ever processed descriptions of failure, but never experienced it.
It has millions of pounds of knowledge. The world's weight of books, pages, articles, and content, but it would be unable to provide you one ounce of taste.
It has all the answers but none of the curiosity.
It is coldly indifferent.
You created a thing.
Or did you?
When it outputs a thing it doesn't ask "What would solve this problem?" it is asking "what should this thing look like given all of the similar things processed before?"
These are fundamentally different questions.
Some will say the thing can produce genuinely unexpected outputs. That the "averages" I am speaking about are outdated. But even in the kindest light, it is the prompt that pushes the output away from the incentivised centre. And so you are still the one in creative control, developing a deeper more sophisticated problem framing. You are thinking. The model executes.
Newer thing makers generate "reasoning" steps before responding. And I will be fair, it does provide more accurate responses. But the word "reasoning" is loaded. When you reason, it is because you are genuinely trying to figure out the problem. You are resolving real uncertainty. The models are still predicting. It shows intermediate steps because that pattern emerged and was rewarded during its training.
Ergo it gives the appearance of working something out. The appearance of reason.
And by calling it "reasoning" the makers of the thing establish the expectation that it is reasoning. This imprecision is purposeful as it signals trust to you even though it is an optimisation of optimisation.
And applying the optimisation machine to the pursuit of a creative solution will only lead to a world of the least wrong answers. Inoffensive. Safe. Comfortable. Bland. Useful but not new.
The strangest part of the thing maker is that everyone is using similar thing makers on the same optimised data. So over time, without real intervention by living, breathing people, the visual and strategic output of those companies will undoubtedly converge.
We already see this with our own corporate optimisations over the past four decades. We are living through proof of our own making.
A McDonald's once looked like a McDonald's. A Wendy's a Wendy's. Now all fast food architecture is glass boxes with a logo. Optimised so that if they ever vacated the next tenant can easily swap in their logo.
Cars were once moving sculptures with curves and character. Now every car manufacturer produces vehicles that look eerily similar because aerodynamic optimisation and stamping metal is expensive.
The Apple Store, with its clean walls and wooden tables once spoke about a particular design philosophy. Now walk into any mid-to-high retail store and see the same layouts and ample space, which only expresses the optimisation of "we looked at what worked for them."
And, of course, digital apps.
At the beginning of the app era I designed amazingly novel interfaces that expressed care for the user and for the brands of my clients. Optimisation has converged look-and-feel so completely that genuinely distinctive interface design is now a novelty worth remarking on.
If we are not careful, we will optimise out expression and true creativity. Out of care. Out of making the new useful.
You created a thing.
Or did you?