To what level do we want AI to participate in creation? Search, Interpretation, or Decision?

Sharing is Caring

The pros and cons of AI

In the past month, I wrote several articles about the impact of AI and ChatGPT on the future creative field, with both positive and negative angles.

On the positive side, AI (generally referred to as various mechanisms that can automatically generate text and video works), which is currently in its infancy, still has a long way to go in the future, and it is likely that it will develop to an unimaginable level in our lifetime; Originally, a lot of knowledge search, data integration and interpretation, and the generation of basic works can be done by AI in the future.

The flipside is that, in addition to potentially displacing some of the existing professions, there are still errors (perhaps from algorithms or raw information), and irresponsible use (such as for writing papers for lazy, or producing fake news ) and so on.

Speaking of “replacement”, what AI can replace is “jobs”, but not necessarily “people”; people can turn to do things that AI can’t do, or higher-level jobs. But as I said earlier :

If you only know the skills that AI can replace, and you are not as good as AI, you may be replaced. Before the 1960s, there was still the job of “elevator operator” in the elevator; in the 1980s, there was also the job of “report printing administrator” in the computer room, but these skills and positions have since disappeared.

If I were to put it in a more vague vernacular, one of the worrying points would be: machines are getting smarter, but people are getting dumber because of it.

As for whether the result is good or bad, everyone can imagine for themselves.

The reason why it is vague is that not everyone will become stupid. Some people even think that although they can’t write articles or draw pictures, they suddenly become good because they can use AI tools. Smarter than ever.

Another point of view is that compared with the era without tools, human brute force is limited, so we invented various tools from hammers to spaceships, allowing humans to have different achievements from animals; and AI is another time The industrial revolution has enabled machines to not only strengthen human brute force, but also extend human intelligence.

There is nothing wrong with that either.

The Five Levels of AI Capabilities

Taking the information age as an example , it has only been about 30 years since machines have been able to help us search data across eras and regions ; The result of the class (think about it, how big a project was 30 years ago), is just the application of statistical algorithms, which is not too difficult.

The next ” interpretation ” is the ability that will appear when AI grows to a certain extent and is connected with a large amount of external data (ChatGPT does not yet have it), such as providing such a report: “The global economy grew by 10% last year, But if there were no wars affecting the supply of energy and raw materials, it should be possible to reach 15%.

So far, it is still at the level that “neutral” data interpretation and comparison can achieve, but the next “cognition” and “decision-making” begin to enter the highlight.

Taking the above example (of course using extremely simplified logic as an example), the ” cognitive ” ability is based on the premise that “war affects economic development” and “positive economic development is beneficial to mankind”, and it is obtained that “a country that launched a war is a negative factor”.

The ” decision-making ” is based on the premise that “a certain country is a negative factor” and makes a strategy of “should launch missiles against a certain country” or “impose sanctions on a certain country”.

This process is not new at all, at least in the world of human analysts; and, if you’ve seen movies like Terminator or the 1983 version of War Games , you’ll know they’re all about AI Errors in the process from “interpretation” to “cognition” lead to wrong decisions and almost catastrophic results.

This is because technically speaking, AI will evolve to the stage of “cognition” and “decision-making” capabilities sooner or later, so we will face several problems:

  1. Can humans hand over these two stages of something to AI? For example, whether to save the wife or the mother first, or whether the train should choose to kill one person or five people?
  2. If so, where would the line be? Or the final decision still has to go through human approval? Are humans infallible?
  3. If it must be recognized by humans who have emotions, must eat and sleep, and will be tired, can the value of AI in terms of efficiency still be exerted?
  4. If humans decide to prohibit AI from intervening in “cognition” and “decision-making” (even if it is technically possible), what kind of knowledge, ability, and training must people have to be able to rank at the top of this pyramid without becoming a group eliminated by AI?

artist’s view

A few days ago I just met some artists. A is a skilled painter who has long been famous for his combination of fine brushwork and semi-abstract techniques; although there are many imitators, no one can surpass it so far.

When I talked to A about the issue of AI automatically generating images, I jokingly said: “With the ability of AI drawing in the future, as far as general-purpose illustrations are concerned, artists will probably only have a problem of “dignity” in the future. You may wish to try it in the future. For the AI, “Please use the style of painter A to generate a work with XX theme”, this may save you a lot of trouble.”

Painter A: “But at present, AI can’t really create new works. Technically, it just follows instructions and uses similar works to transform collages. Maybe it can be made a bit similar on the surface, but the essence still can’t be done; what’s more. There is still a problem with the originality of using other people’s works to make collages.”

Later, when I met artist B, he said: “AI currently has little influence on real artists. Although the collaged works can meet the needs of ordinary people with pictures, these people are not the artist’s customer base; and the collage The more pictures you have, the more you can highlight the value of original works, and the two can be regarded as not conflicting.”

Arriving at a certain exhibition venue in the evening, I met a C artist who mainly created T-shirt patterns and was setting up a stall. He said: “Hey hey hey, look (pointing to the own-brand T-shirts on the booth), the patterns on them are all drawn by AI, which saves me a lot of effort and won’t clash with others.”

C also told me: “In fact, the first ones that will be impacted by AI-generated images are those who “sell image galleries”. In the past, image libraries were very expensive, and most of them did not fully meet the needs, and they had to be synthesized and modified by themselves; AI-generated images only need to be ordered. Yes, the generated pictures can be used immediately, even if it costs 10 US dollars, it is cheaper than buying a single photo from a high-quality library, or combining it yourself. The library is dead, but the AI ​​​​generated picture is alive.”

Back to the “dignity” factor mentioned above. This is not to say that they will reject AI tools for the sake of “the dignity of the artist”, but just as they have accepted new tools in the past (such as drawing from paper and pen to digital drawing tablets, and then tablets), they will use AI tools in the most suitable way. Integrate into the workflow in a unique way, but I still leave my own persistence in the final “style” and “human touch”.

Find out how AI can help your workflow

The same goes for marketing and writing. Leaving aside the aforementioned “irresponsible use” perspective, we also don’t necessarily have to reject AI tools (of course it doesn’t matter if you want to), but to see if it can find a favorable position in your workflow, Improve the “quality”, “quantity”, or both of your work results.

From my understanding, AI will be a very good tool in marketing, including data collection, analysis, interpretation, etc.; because marketing has two major perspectives, one is “human nature” from top to bottom (understanding customer/social needs , problem solving, empathy), and the other is the analysis and interpretation of “data” from the bottom up.

In fact, marketing can also be said to be a kind of data science, and this angle is the strength of computers, the Internet, and AI.

And the “humanity” aspect from top to bottom is like “the dignity of an artist”, or the two levels of “cognition” and “decision-making” mentioned above, all of which must be made based on the achievements of the previous levels Potentially correct and cost-effective decisions:

Should missiles be launched against a certain country to solve global economic problems? Should our products be high-end and exquisite or popular? Should the copywriting be funny, bloody, or warm? Where do I put the last touch of red paint on my work? Does the wife want to save her?

For some things, perhaps AI has the ability to help you decide, but for whatever reason, you may feel that it is better to do it yourself.


I would like to share some sincere thoughts: At present, many people are having a good time and even amazed by the Midjourney drawing and ChatGPT question and answer. In my opinion, the results are mostly “disposable chopsticks” level.

Because, apart from posting something on Facebook and praising it, you usually won’t use it a second time, or even look at it a second time, let alone use it as a reference or achievement for your work; The attitude and amazingness of AI will more or less reflect what kind of people you have seen in this field, or how much effort you have put in.

However, for the AI ​​system, this “training” process is still necessary, and what it spits out to you is just a reward for you to entertain it.

Personally, the current AI tools and finished products are not very helpful, so they have not yet entered my workflow. Still, there’s certainly a chance of becoming an indispensable utility in the future, so I’ll still keep an eye on it.

Friends who have read my articles for a long time know that analyzing the copywriting of various countries and discovering the beauty of it is one of the topics I often write about; the situation I don’t want to encounter is to see a well-written article full of wisdom He Jifeng’s text, after deciphering and admiring it for a long time, he finally found out that it was written by a machine.

But if there is such a day, you can only accept it calmly.

Sharing is Caring