The danger of an answer that arrives too early

A blank Figma file used to have a particular kind of discomfort. So did an empty document, a whiteboard before a workshop or the first few minutes of a meeting where everyone understood that there was a problem but nobody knew quite yet what shape it had.

Nothing was there to react to. You had to stay with it for a while.

Now that space can disappear almost instantly. You can take a brief that is still slightly unclear, paste it into a chat and ask for five possible directions. A few seconds later you have a structure, some hypotheses, potential features, perhaps even a name. The uncomfortable part is over. There is something on the screen.

And because something exists, the conversation can begin.

Someone likes option three. Someone else thinks one and four could be combined. Another person starts pointing out what would not work technically. A designer begins imagining the interface. The team is moving, and from the outside it looks exactly like progress.

But something happened before any of that.

The first answer gave the problem a shape.

I have been thinking about this because I increasingly use AI at the beginning of things, not only at the end. To organise thoughts, look for alternatives, challenge an idea, understand something faster or simply get past the resistance of starting. It is incredibly useful for that. But I have also noticed how quickly an answer, even an average one, can become the centre of gravity of everything that comes afterwards.

Once there is a first draft, we stop facing an empty space and start reacting to what is already there. We improve it, reject parts of it, rearrange it, make it more realistic. We may end somewhere completely different, but that first response has already established a set of boundaries around the conversation.

This is not something AI invented.

A first wireframe can do the same thing. Put one interface on the screen early enough and suddenly a conversation that could have been about the problem becomes a conversation about whether the button should be there. The first number in a negotiation changes what feels expensive or cheap. The first concept that gets visualised properly often receives more attention than ideas that are still difficult to articulate.

What feels different now is how little time has to pass before that first reference point appears.

We can produce an answer before we have experienced much of the uncertainty that used to come before it.

And uncertainty has terrible UX.

It feels slow. It gives us nothing tangible to show. You cannot put “we spent two hours realising we were asking the wrong question” into a roadmap and make it look particularly impressive. A screen, a concept or a list of possibilities feels much more like work has happened.

But quite a lot can happen before there is anything to show.

Someone notices that two people have been using the same word to mean different things. A question exposes an assumption hidden inside the brief. The team realises that the request that arrived as “we need this feature” is actually a symptom of something else. A conversation goes backwards before it can move forwards.

None of those moments produce much.

They can still change an entire project.

Perhaps this is why I am becoming interested in the difference between using AI to explore a problem and using it to give the problem its first shape.

They sound almost identical.

“Give me five ways to solve this” and “what might I be missing about this problem?” can be typed into the same box and answered by the same system, but they invite very different kinds of thinking. One begins producing directions. The other keeps the space open for a little longer.

I don't think one is inherently better. There are plenty of moments when speed is exactly what is needed. Starting from zero every time is neither noble nor particularly efficient, and pretending we should protect the blank page from technology would be absurd.

The thing I do not want to lose is the ability to recognise when the blank page is still useful.

Because sometimes not knowing what the answer is yet is information.

It can mean the problem needs more context. That the question is too narrow. That the people in the room do not agree on what they are trying to achieve. Or simply that an idea needs a little more time before we give it a form.

For a long time, we treated blank space as the obstacle before the work began.

I'm starting to think it was sometimes part of the work.

AI gives us an extraordinary ability to make something appear almost immediately. And I love that ability.

I'm just becoming more careful about the moment in which I use it.

Because the first answer does not have to be particularly good to influence everything that follows.

Sometimes it just has to arrive first.