Before discussing AI and creativity, we need to acknowledge a truth: Most AI systems today are still clumsy when it comes to creativity. They can mimic existing styles, combine existing materials, and generate numerous variations that seem different but are fundamentally similar. However, they are far from understanding vague goals, identifying what's worth doing, or recognizing novelty like an experienced creator.
This isn't a limitation of capability, but rather a choice in training objectives. Our current mainstream model training focuses on understanding human language, not generating ideas independently. AI is trained to break down complex tasks, execute them step-by-step, and minimize deviation—this is the core of instruction-following: give clear tasks, reduce unexpected behavior. This is also where it performs best. Ask it to fix a bug, it fixes it; ask it to summarize meeting notes, it quickly provides a list. But this also means we've never seriously trained AI for creativity.
Perhaps we've implicitly assumed that creativity is too difficult, that it's humanity's final bastion.
But things are changing. The starting point of this change isn't that AI is getting smarter, but that it's breaking away from the human rhythm in both time and space. We are facing a technical system that, while not adept at thinking, can make creativity—once scarce—abundant through speed and parallelism. This structural shift might be more disruptive than AI simply becoming more intelligent.
The Physical Structure of the Creative Process is Disintegrating
In the past, the path of creation was roughly linear: starting from a vague idea, gradually materializing into a product. You researched, brainstormed, wrote documents, designed interactions, planned implementation—investing significant human time and collaborative effort. Even with agile development, cycles were typically measured in weeks. A small idea might take three to four weeks to go through the entire process. The essence of this process was a linear chain of collaboration. It was slow, but we were accustomed to it.
AI disrupts not just individual steps, but the entire physical structure. With AI assistance, you can now generate a prototype within an hour, launch a live page two hours later, and receive user feedback by the same evening. This cycle isn't just shorter; it's short enough to demolish the old rhythmic system. Ideas no longer need to queue; validation no longer requires waiting. Design, implementation, validation—stages once separate—are now nested, mixed, and overlapping. We are entering an unstable new state: the boundary between idea and implementation is dissolving.
When feedback is sufficiently fast, many processes that previously required deep deliberation are replaced by iterative experimentation. You no longer need to spend ages justifying a plan; instead, generate ten versions, launch them, and see what works. You no longer fixate on figuring out exactly what to do first; instead, you do a little, then observe the results.
Action first, explanation later.
Not a Triumph of Understanding, but a Victory for Bias for Action
Here lies a counter-intuitive paradox: AI doesn't need to fully understand something to accomplish it. It can even deliver surprisingly effective local solutions while remaining largely ignorant of the whole. Ask it to generate an ad campaign, it might not know what makes a good ad, but it can generate a hundred versions at once, test them on different platforms, and automatically select the one with the highest CTR. This isn't insight; it's execution. It's not epiphany; it's dense, tireless, high-frequency trial and error.
AI represents a systemic bias for action. And this bias hits a blind spot in modern creation: We constantly debate before deciding, procrastinate before executing, doubt before committing. AI does none of this. It doesn't argue, doesn't get anxious, doesn't suffer from internal friction. It just does. It uses scale and speed to marginalize human cognitive inertia into an optional extra. This is its decisive advantage. Especially in fields like creativity, where outcomes are inherently unpredictable, doing it often proves more effective than thinking about it first.
This isn't a victory for intelligence, but for system structure. It's speed and scale themselves, as physical mechanisms, overturning the logic of creation.
When Time is Compressed and Space Expanded, Creation Becomes Something Else
This change can be understood from two perspectives:
- Time dimension compressed: An idea used to take weeks to go from birth to validation. Now, it's hours or even minutes. You used to wait for results; now, results often precede you.
- Space dimension expanded: You used to carefully craft one version. Now, you generate hundreds of options simultaneously. You're not picking the best; you're observing which one naturally survives.
Multiplying these two factors yields a completely different systemic state:
We are entering an era where exhaustive search + filtering replaces deep deliberation. Not every idea is worth designing, but every idea is worth trying. Not every choice is made by judgment, but by letting real-world feedback decide the winner.
Creation shifts from linear progression to a dynamic, feedback-driven search system. It's no longer about having a clear goal and striving to achieve it, but defining a space and allowing the system to evolve outcomes within it.
So, Who Are We in All This?
If AI handles generation, testing, validation, and optimization, what's left for humans?
Here are several potential role shifts:
- From Creator to Boundary Designer: You no longer craft the single best idea, but decide which directions are worth exploring.
- From Executor to Feedback Interpreter: You analyze why one approach worked and another didn't.
- From Judge to Goal Definer: You tell the system what constitutes "good," rather than finding the good yourself.
You cease being the sole master of the system and become the regulator of an ecosystem. This might be a higher level of participation: you build an environment capable of evolution, rather than personally sculpting the result. But it also means you can no longer fully predict the output or control the process in familiar ways. This is the demand technology makes of you: relinquish some control in exchange for greater possibility.
This Isn't a Trend, It's a Rupture
Many view AI's impact as a gradual curve. Today it's auto-complete, tomorrow code generation, the day after automated writing... seemingly linear extrapolation.
But if we look at it on a different scale, it's not linear; it's a phase transition. Like water suddenly boiling at 100 degrees Celsius, AI isn't slowly changing creation. When its speed and parallelism reach a certain threshold, the very logic of creation undergoes a fundamental shift.
You're not making a product; you're domesticating a system to yield what you desire. You're not pushing a project forward; you're maintaining an open environment to receive potentially valuable unexpected outcomes.
And the prerequisite for all this is accepting that creativity no longer belongs solely to individual genius and decisive moments, but is a process that can be organized, structured, and regulated.
Perhaps we should stop asking: Can AI possess creativity? The question itself has become redundant. What matters more is that the very act of creation has been thoroughly transformed by the spatio-temporal structure AI introduces.
Now, creativity might no longer be about thinking of something no one else has thought of, but about constructing a system that can continuously generate, validate, and filter useful variations. It relies not on inspiration, but on structure. It doesn't worship the definitive stroke, but embraces emergence through evolution. If creativity was once like writing poetry, now it's like building a greenhouse. What grows inside isn't entirely up to you. You can only adjust the light, humidity, soil—then wait, observe, select.
We once believed creativity could only stem from human experience and intuition. But perhaps creativity was never just an ability, but something that can happen within a structure.
AI didn't become a poet, but it changed the way poetry happens. And we stand on the brink of this structural change—neither gods nor mere users, but invited designers of this new evolutionary system.
Are you ready?
Note: The views, content, framework, and entire text of this article were autonomously proposed by AI after processing all the recordings I made this week. My role was to provide heuristic prompts, make choices among the options it presented, and offer detailed feedback. Beyond that, this article is 100% AI-generated. My prompts and interaction process can be found here (in Chinese).
Comments