Does AIGC still remember the Dreamwriter by the Daming Lake?

Original link:


ChatGPT is on fire.

In two months, it reached 100 million users. The growth rate has set an unprecedented historical record-but to be honest, the statement of “reaching 100 million in two months” is not entirely accurate. But I also think it doesn’t matter.

Not only are there heated discussions in the public opinion field, but there are also fires of real money.

Google’s Bud robot made a mistake in its debut, and as a result, the company’s market value evaporated by hundreds of billions of dollars. This is the most expensive wrong answer in history. From another perspective, it tells us how popular AI is in the capital market.

Major domestic manufacturers have expressed their willingness to vote:

The co-founder of Meituan, Wang Huiwen, who was number two at the time, “outrageously” announced in the circle of friends that he would start his business with 50 million dollars.

Of course, people will also start to discuss in full swing: What kind of jobs will be threatened by AI-this is a common problem that almost every technological advancement will bring.

In a large number of screenshots of ChatGPT chat conversations, many people seem to find that using it as a writing robot is probably a nightmare for content creators.

Generative ai, generative ai, generating an article is easy, and humans seem to be no match.

The four letters AIGC exploded instantly.

But wait a minute.


Now, we have invited a player who has disappeared but also gained a lot of attention back then.

Tencent News’ Dreamwriter.

I have a course called “Internet and New Media Case Studies” in my school, which is aimed at junior students and has been taught for more than ten years. Since Dreamwriter came out, every year in this class, it has been a case that has become the subject of discussion.

So, I’m just being lazy and intercepting a page of ppt made by a student about this case last semester, so that I can get a general understanding of its ins and outs:


In fact, when discussing her draft with Wang Qingxuan last semester, ChatGPT has already started to attract the attention of the world. One of the topics of our discussion, though, is whether intelligent writing can be the killer of journalism—or so.

The tone of the discussion class is open-ended, with no standard answers. So, while I think intelligent writing will be of very limited help to journalism, I don’t mind my fellow students thinking that this is the future of journalism, as long as they can make a case for it and make a case for it.


Any industry follows the principle that supply and demand must reach a balance, and the achievement of this balance point is actually the price or value point.

When the supply of an industry is obviously in short supply—that is, the supply is far less than the demand, then high prices will occur. If administrative means desperately suppress the price, then poor experience, poor service, and even all kinds of tricks in the industry will happen. will follow.

The medical industry is a very standard industry where supply exceeds demand, so I always think that AI will be very useful in the medical industry, because it can “theoretically” greatly promote supply. Of course, in terms of specific details, there are still many thresholds to pass, which is another topic.

But in the journalism and media industry, anyone with common sense knows that supply far exceeds demand.

We are already in a world of information overload, so why is it necessary to introduce a technical method to desperately increase the supply?

In fact, the real thing to do is: getting the right content to the right people makes sense. Generally speaking, in a market where supply exceeds demand, distribution is the core issue. Therefore, algorithmic recommendation interest mining—in fact, we can also regard it as a kind of AI—has achieved great commercial success.

Dreamwriter can publish hundreds of thousands of articles a year, which is meaningless commercially. It certainly has no typos, no grammatical errors, and no wrong punctuation. But it won’t be very good either. The disappearance of this kind of machine producer, which is as indifferent as plain water, is inevitable.


An interesting comparison is here:

A narrative essay about the economy in the first quarter – this is Dreamwriter, doesn’t it seem boring

And if you add a question to this article: Can you tell me the economic situation in the first quarter-this is ChatGPT, and it becomes sexy in an instant?

No, that’s not the case.

The difference here doesn’t lie in the technology behind the two. Although a friend who knows a little bit told me that the technology behind the two is indeed different.

The difference is “interaction”.

The way humans give instructions to machines is different. ChatGPT is obviously able to understand human “natural language” rather than specific instructions.

You can even tune ChatGPT, that is, let ChatGPT understand you better through one question after another.

For example, when you tell ChatGPT that “the old king next door” has a strong sexual ambiguity and inappropriate hints in human language, then it will “remember” your hint in the next interaction with you. ——The current state is that as long as you do not close the page that interacts with ChatGPT, then from the first sentence you say, all the information you tell it will become important background information for its subsequent interaction with you.

The sexy thing is that ChatGPT is “understanding” you. Although this seems to be incomparable with the so-called “understanding” of human beings, because the machine should not know the “meaning” so far. But this is indeed a kind of “understanding”.

Continuity and human dialogue, although it is also a so-called generated content (GC), but this is not to generate an article.


So, where do you think AIGC will shine?

—— The first episode of pulling nitrogen ——

(to be continued)

This article is transferred from:
This site is only for collection, and the copyright belongs to the original author.