A few days ago, people in an AI group were discussing whether to pay for a structured course in AI video.

Someone who had just started said they had spent eight hours trying to generate a single scene and still hadn't got it right. They were wondering whether they were going about it the wrong way. Someone else suggested signing up for a course: spending money buys you time.

I shared my approach too. I've never taken a structured course in AI video. I simply talk to AI about the scene I have in mind, have it generate a script, and then take that to the platform to produce prompts. There is plenty of trial and error, plenty of adjusting, and quite a bit of wrestling with it. But I think the finished videos look pretty good.

As for paying to learn, I spend quite a lot on it every year. I've taken an illustration course with my friend Tian Xiaozhuang, and I belong to learning communities run by Liu Run, Wang Runyu, and others. Paying to save time is something I've long done, and I'll keep doing it. So what follows comes from being part of this world, rather than looking in from outside.

One question, several answers. The discussion was lively, but I've thought about this for quite a while, and my answer is actually quite direct. Methods alone are not worth spending a lot of money to learn. There are things worth paying for, though. They just aren't where you might expect.

Why methods aren't worth big money

In the AI era, methods are everywhere. If you can speak, type, and explain what you need clearly, the content comes out. That goes for images, writing, and videos alike. Precisely because it comes so easily, it is no longer scarce in itself.

Scarcity creates value.

But I need to add something: being plentiful doesn't mean being useless.

The value has simply moved somewhere else. What used to be "I'll explain it to you" has become "I'll help you choose and verify it." That sounds roundabout, but it is very practical. You first have to know what approaches exist before you can know where to put your effort. Without that knowledge, you don't even have a direction to work toward.

So people who say "methods have no value" are only half right. They aren't worth big money, but their value isn't zero either.

What is worth paying for

Then what is worth paying for? My criterion is very simple. I look past what someone says and ask whether they have actually delivered work.

Someone in the group put it rather well: the biggest money in this industry comes from information gaps, not technical skill. If someone's technical work were truly at the very highest level, they would already be making major films, rather than sitting here teaching you.

I can speak to that from experience. More than ten years ago, I sold educational content and courses on Weibo and reached seven figures for several years running. Looking back, much of that came from arriving at the right time. The market wasn't this competitive. Many people wanted to learn, and few could teach. The demand was right there. I followed it and built things up little by little.

In other words, my methods were valuable back then largely because they were scarce. Today, that scarcity has gone, so the same methods are no longer worth big money. It isn't that anyone has got worse. The conditions have changed.

There isn't much point in only revisiting the past, so let me talk about something more recent. I make AI videos without ever having taken a structured course. I describe the scenes in my head to AI, ask it for scripts and prompts, and work them out through repeated trials and adjustments. My "Ask Runyu" series won recognition from Runyu, his team, and some members of his community. Later, I also completed paid commissions for custom creative AI videos.

Delivering that work mattered to me. It lets me say these things without feeling like a fraud. I've been in the water and swum through it myself. I know how deep it is and which holes really can twist your ankle.

Methods and delivery are two different things

So I look beyond the label of "course" or "institution" to the person behind it.

Some courses make the methods sound wonderful, but then stop explaining just before the final step that would get the work delivered. The instructor may not be deliberately holding something back. They may simply never have completed that step themselves. I've also seen instructors with a long list of achievements in their profiles. Look closely, though, and those achievements turn out to be closer to a buzz funded by course fees, such as view counts built by pouring money into advertising. Ask how much they spent and whether they made it back, and they can't answer. The numbers look good, but they don't translate into anything you can use.

The other kind is one I respect: people who provide ongoing coaching, help develop a business, and offer resources and access to channels. These usually cost more, but the people providing them are making money themselves, and their business models have been shaped by the market. Part of what you pay buys access, at a fraction of the cost, to experience they actually earned. That is worth it.

Ultimately, methods and delivery are two different things. Methods can be sold wholesale. Delivery has to happen one job at a time.

Being able to make something is still a whole way of thinking away from earning money

I know some people will object: if I don't take a course, can't I figure it out myself? You can. But being able to make something still leaves you a long way from earning money.

Here's an example. Some teams making AI comic dramas have brought production costs down from around 30,000 yuan to a few thousand. By ordinary logic, they should get to keep the savings. What happened? The content became cheaper to produce, but buying a view became more expensive. All the savings went back into buying traffic, sometimes with extra money on top. The people using the latest tools ended up becoming more efficient workers for the platforms.

Between being able to make something and being able to earn money lies a whole way of thinking, not a single tool.

Then there is the wider picture. The supply of content is unlimited, but attention is finite. AI has cut the cost of producing content by more than half, and the number of AI comic dramas being made has multiplied dozens of times in a year. Yet an explosion in supply doesn't mean everyone gets a share. Seen from the other side, being noticed has itself become increasingly scarce. Someone in the group mentioned a creator who taught himself for a few months, reached 100,000 followers, and now earns money by taking on students. His advantage was that he understood early on how content spreads. Whether something takes off has long been about more than technical skill. It is even more a question of communication and distribution.

What AI cannot replace is judgment

Here's an example. Once, when I asked AI to research something, it solemnly told me that Tee Sang had no website. I laughed when I saw it, because I knew perfectly well that I had my own website. It sounded thoroughly convincing, but it was false. You spot an error like that at a glance because you know the subject yourself. At other times, it isn't so obvious. The information it gives you all looks right, and then you have to cross-check it with several models and platforms before you can be sure.

But that isn't the only kind of judgment. There is also judging the timing, which is harder.

Take AI video. A friend spoke to me around May or June. He was working at an AI video company that happened to be moving into overseas business. He caught that moment and became one of the pioneers of its overseas operation, leading around a hundred people and putting out more than a hundred videos a week. He reviewed the work and kept an eye on overseas markets, earning over 100,000 but under 200,000 yuan a month. He had only been learning AI video for a little over a month. Getting into that position owed a great deal to the circumstances and opportunity coming together. Not everyone can repeat his experience, but the money he earned was real. Now they teach courses. The stories they tell are real, and so are their credentials.

But the market has already changed. The payouts offered by platforms have kept falling. The revenue from the same piece of content is much lower than it was at the peak, while more and more people keep entering. Most coaches won't bring these things up on their own. They may not be trying to deceive you. They are talking about their experience at that time, while you are facing a different market.

And you are still wondering whether you can earn as much as he did. That is where your own judgment comes in.

It is true that he made money, and it is also true that that market has passed. Both things can be true at once.

AI can give you ten plans in a few seconds. But which one is right, which should be dropped, and which would go wrong if you started now? It doesn't know. You have to know. Someone in the group said technical skills could be refined later, while what really needed developing was aesthetic judgment, creativity, and imagination. I agree. With the technical part, learning it six months later doesn't matter much. AI can already control your computer and operate software directly. Things that once took ages to learn may eventually take just one sentence. Judgment has to grow through doing the work yourself, one attempt at a time, and reviewing it again and again.

So what really has value often lies beyond "how to do it," in "when to do it, which thing to do, and when to stop." Nobody else can do that for you.

What I want to do

Once I thought this through, I actually felt more at ease. Next, I want to hold a session sharing what I've learned about GEO. Put simply, GEO is about getting your content recommended preferentially in AI search, a new approach following SEO. I'll share the methods for free, explaining how I understand this field and holding nothing back. But I will charge for ongoing guidance and reviewing the work together.

It isn't because I want to make that little bit of money. After all these years of teaching, I've come to understand more and more that teaching and learning need a commitment on both sides. I've seen too many people pay tuition and then never study. I don't even feel comfortable taking that money. On the other hand, people who pay are the ones who actually get started, while most people who come for free just listen. That payment filters for the people who really intend to follow through, and makes the arrangement more responsible on both sides.

Anyone can talk about methods, and I'll share mine for free. Judgment and support have a price. What I give away is what everyone already has. What I charge for is the part nobody else can do for you.

There, I've said it. Courtesies first, ground rules second.

Finally, I'd like to hear from you. Does this make sense, or do you see it differently? Have you paid to learn AI recently? What did you spend the money on, and looking back, was it worth it? Tell me in the comments. I'll read every one.