Text | Li Zhiyong
From the perspective of products, technology generally looks at the effect without looking at the principle.
Every day I take a moment to observe the latest advances in AI. The method is to have an API, the main method is to build a test set of various tests, and I mainly look at what the unofficial production is like, and whether it is universal.
After watching more of the above videos, I have two thoughts:
First, it's time for video generation to explode.
Second, there is a high probability that GPT5 will be able to pass, although it is difficult to say whether OpenAI will pass first.
Between this new point of opportunity and the present, it is actually a small low tide of AI.
Why is video generation going to explode?
The official video of the model party is because of the stakes involved, at most you can see the signs, and it is not very credible.
So you have to change the perspective, if the average user continues to explode wonderful videos with a certain frequency and magnitude, it will almost reach a critical point. You see, MJ was like this before it became popular, and all kinds of pictures continued to be exposed. Because the continuous explosion of videos on social networks must meet two conditions:
The first one is surprisingly quality. Beyond the original understanding.
The second one has to be relatively easy and inexpensive, and many people can do it.
The former is the timing, the latter is the foundation. Not one is missing. Now there is no one who has nothing to do to explode the picture, even if the picture is more amazing than what was done in the past, no one looks at it.
Now it looks like these two conditions are basically met, just a little bit. Once this is satisfied, it becomes the next mass opportunity point for AI.
Why mass opportunity?
AIGV (Video) is a word I coined apocryply, but it is true that a distinction should be made between generated text, images, and videos. Because they can't be the same magnitude of influence. The income of domestic head anchors is not in the same order of magnitude as the income of bloggers in the past, and it is not even ten times and a hundred times the difference. If Douyin is listed, how many Xiaohongshu and Zhihu are equal?
Now this video ecology is facing a reshuffle, one wave of people is going to machine guns, and the other wave of people is still big knives and spears, so I am afraid that the impact will be much greater than that of past article generation. He will push the video platform in an unpredictable direction.
On the contrary, we can see the part outside the video platform more clearly, and even the development route has been practiced in the past.
First of all, the new version of Li Yizhou will come out in waves. At this time, content generation faces dual specialization, part of which is the model of video generation, and part of which is some skills of the video itself.
There will still be tools in the shell, Heygen, a tool that doesn't completely produce video, is $20 million a year, and the new tool here should be worth more.
A lot of AI-generated content then makes its way to various platforms. It is estimated that each platform will be a little tangled, but in the end, it will be like now, basically allowed, at most add a mark.
In the future, it will develop in depth, not only at the level of promotional films, but also to short dramas (jokes, animations, etc.). At this time, it will blow up again, because it overlaps with the progress of the previous wave. On the AI collision game before, Zhou Wei shared his Agent who writes short stories, and after listening to it, you will know that this kind of product is combined with video generation capabilities, which is a special fit. The value of such products will suddenly rise a lot.
After that, it is estimated that it will be a long drama and a live broadcast. Among them, the live broadcast is a little more complicated, because it has to be combined with the often said motion capture, and it may take some time to make it particularly realistic.
In the end, it must be a movie. What does it feel like to throw a novel, AI rewrite the script, and AI generate a movie?
AI first vertical and then horizontal
As mentioned several times in previous articles, the biggest difference between AI and the Internet is that it is first vertical and then horizontal, like digging a deep well, and now there is one more deep well here.
Ventu is a well, video is a well, and it is deeper and bigger.
What's even more interesting is that this is a mass opportunity, and it seems that big manufacturers can't do anything to monopolize products based on this.
Everyone is almost equal, it assists in the realization of your ideas, and cultivates a group of tools and teaching tools.
First vertical and then horizontal will lead to such a state: it will not blow up so badly all at once, such as a WeChat all at once, but it will continue to blow up.
At this time, the basic methodology for facing it is not far-sighted at all, but to return to the scene, it is necessary to gather cases such as AI collision bureaus, so that everyone can vividly feel the means of AI application on the scene.
To put it simply, it is to seek truth from facts, you have to know what the situation is, what others have done, and then form your own judgment, what you don't understand should not exist, at least not to do things.
It's very different from the internet. It was a big wave that rushed to the end and passed.
What does it look like to be thoroughly matured?
I posted such a video before, but I didn't expect that there were a few experts in this field in my readership group (let's send a private message to the readership), and after a few more conversations, I found that this is a mature industry.
As soon as the above generative video technology matures, coupled with this set of technologies in the video, it is possible to give birth to this kind of product:
A person has a doppelgänger of their own, and this avatar is the kind of product in the video, and then the script, the content is generated, and the release is fully automated, and you can publish the content by saying it. I don't want to cut the film or something. This market should still be very large, and those who are really willing to do it can try.
This thing only has one more mobile phone per person, but this mobile phone can move on its own, generate content and publish content, which is actually similar to a live broadcast rack, microphone or something, and the amount required is still very large. Some people estimate that it is used to be a water army, but it is illegal to be a water army, just like someone using a mobile phone to make a fraudulent call, but it cannot be said that the problem is similar to the phone, and drones are not also used to throw bombs.
GPT5 will definitely come eventually
Looking at the progress of the video, I am more certain that GPT5 will definitely come.
Now GPT5 is a symbol, and I can't tell what it is, the metaphor is that AI can go one step further. It's not necessarily the next version of OpenAI, it marks whether general intelligence can jump another level, and the impact of each jump in the versatility of intelligence is definitely different from simply generating videos.
Why is GPT5 coming?
For example, if you understand that the physical world has your own model of the world, it is not critical to understand the physical world when generating text and images. But if you don't understand the physical world when generating videos, you can't do the subtleties well.
Explain why you can read the paper, but the result is that AI's understanding of the physical world is deepening, and it is forming its own model of the world.
If one day AIGV (Video) can generate all kinds of TV series in the real world, as well as all kinds of wonderful fantasy movies, then its world model will inevitably cover both the real space and the virtual world (such as the Marvel Universe).
From this point of view, video generation is more or less synchronous with GPT5.
It's hard to say whether it's OpenAI or not, OpenAI looks a little blind after a fight, and it's always very casual to send products, and then Sam Altman doesn't solve the company's imminent problems, and has nothing to do to make political-related remarks, which is not the right way to look at it. But that's another story.
brief summary
When you think of the ultimate point of AI, you may think of yourself with nowhere to put it and panic, but looking at the process, it is actually creating mass opportunities. At this point in time, there are not as many fields that are full of activity and give opportunities to ordinary people. Or should there be more hugs and collisions. If the level of solidification was 1 before, then the world after AI may be 100 (in the bad ending I predicted), so this may be the last chance in the bad context.