From 2025 to today, the pace of AI-generated video has far exceeded anything we imagined — even "the Moore's Law era" feels too slow a way to describe it. We once thought the journey of generative video from a toy on a desktop to real-world application would be comparable to the leap from 2G to 4G in mobile phones: we could imagine the changes it would bring, but never imagined it would arrive so fast. Things that once required professionals, professional equipment and years of training are becoming easier and easier. Writing articles, writing code, generating images, generating music, generating video — even generating entire stories and characters. All of it seems to proclaim: "the future is already here."

As a colorist of fourteen years, I'm actually no stranger to technological progress. My career itself has unfolded alongside the development of digital imaging — from film processing, digital cinematography and digital intermediate, to DaVinci Resolve and Baselight, then ACES, HDR, color management and display calibration. Throughout my career I've been doing one thing: understanding technology, then using it to realize creative work. So I'm not someone who opposes new technology. Quite the opposite — I was probably among the earlier adopters, pushing technological progress so it could drive better images, better creative work and a better experience. Back then I was one of the "front wave" of that era. The new technology of the past solved one problem: "what we couldn't do before, we can do now — and we can do it better." AI, however, is gradually approaching a different question:

"If more and more things can be done, then why should we still do them — and how?"

That may be the real question AI brings.

Is AI just a tool?

On AI — from the birth of AlphaGo to today's generative video and embodied intelligence, from dinner-table conversation to university lectures — I keep hearing two diametrically opposed views:

One holds that AI is omnipotent. As long as the models are powerful enough, the compute sufficient and the data rich enough, any capability that once took humans years of training to master can ultimately be mastered by machines.

The other holds that AI is just a tool. However advanced a tool may be, it is still a tool: a camera won't replace a director, grading software won't replace a colorist, and of course AI won't replace creators.

I used to strongly agree with the second view. But now I'm starting to think both views may be too one-sided, too arbitrary — because if AI truly comes to possess enough compute and capability, then "AI is just a tool" itself becomes a question worth reopening. Suppose one day AI can understand cinematography, editing, color grading, music, performance, story structure — even human emotion. It can generate the actors, the sets, the camera movement, the lighting, the music, the visual effects and the final image on its own. Can we still simply say, "AI is just a tool; the creator is what matters"? I don't think so. Because once the tool itself begins to possess ever-stronger powers of recursion and classification, the boundary between tool and creator starts to blur. I use the phrase "recursive classification" deliberately, because in essence machine learning can only accomplish "recursion and classification" — but isn't human learning, too, a "recursion and classification" of the stories you've lived through, the knowledge you've acquired, the people you've met?

If you conclude from this that AI will ultimately solve everything, I disagree just as strongly — because capability and purpose are two different things. We often discuss what AI can do: can it write code, can it make a film, can it make music, can it generate a real person, can it generate a world that doesn't exist.

But few people ask: why does it do it?

If you ask AI to generate a film or a short, of course it can produce a beautiful image — beautiful cinematography, beautiful lighting, beautiful color. But why this image? Why is this person here? Why should this shot hold for three seconds rather than four? Why doesn't this character say that line in the end? These aren't questions of "can it be done" but of "why do it." A character is more than appearance and dialogue — there's tone, context, subtext. So I've long held that technology solves "how," while the creator decides "what."

But AI is changing that relationship, because AI has begun to move from "how" into "what." It can not only execute commands, but propose options, revise options, compare options — even generate things we never explicitly asked for.

So the real question of the future may not be whether AI can replace a given profession, but:

When "how" becomes ever cheaper, how much is "what" still worth? And when AI can solve "what," what is left for "us"?

We used to understand a creator as someone who "produces an answer." A director makes a film, a writer writes a book, a musician composes a piece.

But perhaps creation has another meaning:

What a creator really needs to do may be to pose a question.

Truly good work sometimes doesn't give you the answer; it simply makes you start wondering: why would this person do that? Why is the world like this? So the creator in the age of AI may look less and less like a "production person" in the traditional sense, and more like someone who asks questions.

The Meaning of the Creator

Humans have a very strange trait: we do a great many unnecessary things. Someone could clearly stay at home, yet drives dozens of kilometers to climb a mountain; someone knows perfectly well that a song can't solve any practical problem, yet spends months or even years learning an instrument; a director knows perfectly well that the world already has countless love stories, yet still wants to make another one; a colorist knows perfectly well that the audience may never notice a very subtle color shift in a certain shot, yet is still willing to spend hours adjusting it. Judged purely by efficiency, much of this is unnecessary — and yet humans do exactly these things.

We write poems, paint, make films, compose, travel, fall in love, commemorate those who have left us, and waste years of our lives on a dream no one else understands. From the standpoint of survival efficiency, many of these behaviors are even superfluous. But I increasingly feel that it may be precisely because humans pursue these unnecessary things that meaning arises — and that is exactly what AI lacks.

As AI continues to develop, production capability will inevitably become cheaper and cheaper, and the cost of learning lower and lower. If AI can let an ordinary person realize an imagination that once only the large film industry could realize, I think that in itself is a wonderful thing.

Then the real question may become:

When everyone can create, what is worth creating? If in the future AI can accomplish everything meaningful, what is left for humans?

Perhaps precisely those "unnecessary" things.

They are unnecessary, but they matter — because meaning doesn't necessarily come from efficiency.

Sometimes meaning comes precisely from this: I clearly didn't need to do it, but I decided to do it anyway.

This may be the one thing most worth re-understanding and rethinking in the age of AI.

Some Thoughts on Creators

When creation becomes this easy, what do people still want to create? What is worth creating, what is worth keeping, what is worth remembering, what is worth someone spending years of their life to complete? The greatest difference between humans and AI lies not in understanding, not in creation, not in emotion, not in consciousness itself, but in humans' awareness of the self, their pursuit of meaningless things, and their feeling for life.

Returning to the two views above: if AI can one day become a competent creator, then humans should be the ones who feel content — who feel and convey — for that is what distinguishes us from machines and tools. And if AI is merely a tool, then humans should make better use of these tools to express those qualities unique to humanity.

Sadly, we are using AI to mass-produce short videos and audiobooks; through a tool that may replace us, we are assimilating ourselves. We are gradually compressing what is uniquely human — conveyance and pursuit — into information. If the content we consume every day shifts entirely from process to purpose, from appreciation to information acquisition, then what difference does it make whether the creator is human or machine?

I can't judge whether AI's future will be that of a tool or a creator, but lately I've been thinking: perhaps this question is far deeper than the question itself. Much of the philosophical thinking among scholars about banning euthanasia stems from the belief that life should be a process rather than a purpose. I think this precisely illustrates that humanity's pursuit of the meaning of the "meaningless" is exactly what should guide us in how to use tools, how to coexist with AI, and what content to create.

From digital cinema to HDR, I've always been an advocate of new technology. But AI is not a digital cinema camera, not HDR. The arrival of AI can't even be simply regarded as a tool revolution — it deserves far more thought. These reflections determine whether we are spectators or creators; they determine whether we become the "front wave" or the "back wave" of the AI era; and they determine what kind of content we create, what stories we tell, and what we leave behind for those who come after.

Colorist — Wang Bo

2026.08.09