Something strange is happening to our definition of good work.
Every day I see things that would have impressed me enormously a few years ago. A beautifully designed landing page. A polished prototype. A working product. A cinematic video.
And increasingly, the remarkable part of the story is how little time it took to make.
Built in an afternoon. Made in 20 minutes. One prompt. One person doing what would previously have required a team.
And to be clear: I find this incredible.
I've spent most of my career building products, brands and digital experiences. I know what these things used to cost. Not just in money, but in coordination, expertise and time.
Things that required days now take hours. Things that required teams can sometimes be done by one person. Ideas I would once have discarded because the implementation wasn't worth the effort can now be tested before lunch.
That is a profound change. But I think there is a second-order effect we haven't fully internalized yet.
AI compresses the time it takes to reach yesterday's standard. That doesn't make speed the new standard. It makes yesterday's standard worthless.
Not literally worthless, of course. A good product still needs to work. A good website still needs to communicate. A good interface still needs to be usable.
But the economic and creative value of simply reaching that level is collapsing. And that changes what we should be aiming for.
What happens to the standard?
Imagine a project that would have taken two weeks three years ago. Research. Concepts. Design. Iteration. Development. Testing. Refinement.
Today, with the right experience and the right AI tools, perhaps you can reach a comparable result in two days.
The obvious benefit is that you just saved eight days. You can ship earlier. Take on more work. Reduce your costs. Spend more time with your family. All of those are legitimate consequences.
But there is another possibility I find much more interesting: what could you make if you still gave it two weeks?
Most of the conversation around AI productivity is still about compression. How do I write this faster? How do I design this faster? How do I build this faster?
Useful questions. But they assume the destination stays the same.
What if it doesn't? What happens when we take some of the time AI gives us and put it back into the work? Not to make more of it. To make something we previously wouldn't have made at all.
We created the average
The tools available to us have always shaped the things we create. Digital design is a good example.
There was a period when interfaces were extraordinarily detailed. Skeuomorphic interfaces used texture, depth, light, shadow and physical metaphors. Creating them was painstaking.
Then the environment changed. Mobile became dominant. Products became larger. Interfaces needed to work across more devices and platforms. Teams needed reusable systems. Products needed to ship and evolve faster.
Flat design wasn't caused by one constraint. It emerged from a much larger technological and cultural shift.
And a lot of good things came from it. We learned to reduce. We built component systems. We created common interaction patterns. We made increasingly complex software understandable to increasingly large audiences.
But there was a side effect.
We became extremely good at producing a certain kind of good.
Look at enough modern SaaS products and something becomes obvious. Most of them are not badly designed. Quite the opposite. They are remarkably competent.
The typography works. The spacing works. The hierarchy works. The interactions make sense. And yet it can be surprisingly difficult to remember which one you were looking at five minutes later.
We spent years codifying what good digital design looks like. We turned judgment into patterns. Patterns into components. Components into libraries. Libraries into systems.
And then we handed that world to AI.
Which is why I think some of the conversation around AI slop misses something important. We talk about AI slop as if AI suddenly introduced generic design into the world.
I'm not sure it did. AI slop looks generic because we gave AI a generic world to learn from.
It learned our landing pages. Our dashboards. Our typography. Our design systems. Our product conventions. Our taste.
We spent years making the recipe incredibly clear. AI just learned how to cook it.
Good is getting cheap
This is where the implications become more interesting than the tools themselves. AI makes the existing standard extraordinarily cheap. Humans therefore have less economic value in reproducing the existing standard.
And that applies beyond design. If everyone can produce competent copy, visuals, code, research and prototypes dramatically faster, competence doesn't disappear. It becomes the entry ticket.
The baseline rises. What was impressive becomes expected. What was expensive becomes accessible.
And when the cost of reaching the current standard collapses, we can use that efficiency in two ways. We can produce more of the existing standard. Or we can use it to move the standard.
We will obviously do both. But creatively, I'm much more interested in the second.
The cost of curiosity has fallen
For years, there were countless things we simply didn't do because they weren't worth the effort. Details that would never survive prioritization. Experiments too expensive to validate. Ideas that required too much engineering. Entire directions that died with some version of: that's cool, but it's too much work.
Those constraints didn't just determine how long things took. They shaped what we allowed ourselves to imagine.
I notice this in my own work. Before, if I wanted a particular interaction, movement or behavior, I had to decide whether it was important enough to explain, design, prototype, hand off, implement, review and iterate. Every detail had a cost. Enough small costs and eventually you learn to stop asking for certain things. You become pragmatic. Often correctly.
Today I can explore many of those details directly. I can try something. Feel it. Reject it. Change it. Try something else.
The cost of curiosity has fallen. And I think that may ultimately matter more than the cost of production falling.
Because removing production constraints should eventually change more than production. It should change conception.
Yet I see far more evidence that AI has changed how quickly we can create things than evidence that it has changed what we think is worth creating.
That's not an accusation. It makes sense. New technologies often begin by imitating the systems they replace. Early cars looked like carriages. Early websites borrowed from print. Early smartphone interfaces borrowed heavily from physical objects.
Right now, much of AI-assisted creation is still using radically new capabilities to produce very familiar outcomes. That is probably a phase. But it's a phase worth becoming conscious of.
The question I increasingly find myself asking is no longer: how quickly can I finish this? It is: what am I no longer prevented from trying?
What comes after good?
I don't know what the next dominant design language will look like. I hope nobody does. That's the interesting part.
I don't think the answer is simply more visual complexity, more animation or a return to skeuomorphism. That would just create another recipe for AI to learn.
The opportunity is deeper. The relationship between designer and code is changing. Implementation costs are collapsing. The number of iterations and directions we can afford to explore is changing.
Some constraints remain, of course. Human attention remains finite. Cognitive load remains real. Accessibility and usability matter. Good judgment matters.
But many constraints that quietly shaped the aesthetics and economics of digital products are weakening. It would be strange if the products themselves didn't eventually change as a result.
And this changes what it means to be good at making them. If AI can increasingly reproduce what we currently consider good, then reproducing good cannot remain the thing that makes us valuable.
And I don't think the comforting answer is: humans are creative, AI isn't. I wouldn't bet my career on that assumption. AI will get better. The generic baseline will keep moving upward.
So assume AI becomes exceptionally good at reproducing what we consider good today. Then what?
I think part of our job moves upstream. Toward judgment. Toward taste. Toward seeing cultural and technological shifts while they are happening. Toward connecting things that don't yet obviously belong together. Toward deciding what deserves to exist before there is a pattern for it. Toward creating the references that tomorrow's models will eventually learn from.
AI can make the existing standard extraordinarily cheap. Our response shouldn't be to defend the price of producing it. It should be to create what the new standard becomes.
Move the standard
Speed is exciting. There is something almost absurd about watching work that once consumed weeks happen in an afternoon.
But I don't think speed itself will remain much of an advantage. Everyone is getting the same tools. Models will get faster. Implementation will get cheaper. The time advantage will normalize.
What remains interesting is what we do with it.
Maybe you use AI to finish at 2 p.m. instead of 8 p.m. Great. Maybe you use it to run a company with five people instead of fifty. Also great.
But if your ambition is to make exceptional things, there is another option. Take some of the time back. Put it into exploration. Into iteration. Into the details that never survived the old economics. Into directions that previously seemed unreasonable. Use it to question the patterns we have spent the last decade perfecting.
Because the opportunity AI gives us isn't simply to arrive at the old destination faster. It is to ask whether we should still be heading there at all.
AI compresses the time it takes to reach yesterday's standard. The interesting question is what we build after we get there.
Move the standard.