Some divergent thinking after in-depth use of ChatGPT

foreword

Recently, I am trying to use ChatGPT in depth. It is obvious that the improvement of work and life is obvious. ChatGPT is a super player in summarizing and organizing documents. I only need to send articles to it, and it can give feedback within seconds Give me the first draft, if you are not satisfied, you can make it regenerate or continue to optimize. When buying vegetables in the supermarket, you can use it to simply check how many ways this dish can be made. For some cold knowledge or unknown fields in daily life, let it give a brief introduction, so that we can have a basic understanding of this field.

A few days ago, OpenAi released a paper on the potential impact of large language models on the labor market: ” GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models “, the conclusion of the paper pointed out that for about 80% of For Americans, at least 10% of their work tasks will be affected by large models; in addition, there are about 19% of people whose work tasks are affected by more than 50% .

In terms of specific occupations, occupations with a higher degree of impact include tax preparers, writers, mathematicians, web designers and journalists .

The groups that are not at all affected by large models are mainly manual workers, including cooks, dishwashers, floor workers, and so on.

The impact in this paper does not mean that such workers will be replaced or optimized, but that Ai can reduce the workload of these workers by a corresponding proportion. Various tasks are proposed by humans, Ai tries to fill this task, and finally it is optimized by humans.

For example, this article, the main content of this article is about some thoughts on the daily use of ChatGPT, and its future prospects. In addition, this article is partially generated by ChatGPT.

The final form of artificial intelligence

Everyone has their own opinion on this question.

But for me, the Auto-Pilot driving the Axiom in Wall-E is the ultimate form of artificial intelligence. The ecological balance inside the spacecraft provides services such as security, food, water and medical care for humans on the spacecraft.

In the movie, it is precisely because its safety control awareness is too strong that the autopilot system prevents humans from leaving the spaceship, and while controlling the entire ship, it makes humans lose the opportunity for autonomous control.

But for us now, all this is too far away, and no matter how ChatGPT evolves, it is impossible to upgrade to the level of Auto-Pilot in the movie. But there is a segment in this movie that ChatGPT is fully capable of,

【Video on YouTube】

ChatGPT and search engines

What is shown in the above video is a function of the Auto-Pilot subsystem. According to the voice input and output, the explanation of the corresponding vocabulary is not what ChatGPT can do at this stage. By the way, because I haven’t watched this movie for a long time, it took me about two minutes to cut this part of the clip. I first asked ChatGPT where this episode was, and then jumped directly to record it.

Although ChatGPT currently has a variety of questions, and the correctness of the answers needs to be improved, as time goes by, ChatGPT’s feedback will become faster and more accurate. At that time, why do we still need a search engine?

So Microsoft directly adds it to its various products, including but not limited to bing, office, Windows, etc. Although Google made a big joke when it launched Bard, it has recently reopened the internal beta. Domestic Baidu also released a statement, although there is no open test, but as far as the press conference is concerned, there is a gap with CahtGPT.

future

During this time, I revisited an old book, “The Algorithm of Love” by Liu Yukun, which is a collection of science fiction short stories. Some interesting chapters in it became reality a few years after the article was published.

For example, the “Master of Art” in the book: It tells about a little girl who loves film and television art and gets an interview opportunity by researching and remaking the film created by her favorite film and television company [Signal Light]. However, after the interview, I found out that all the films were produced by an artificial intelligence called “Old Sammy”, and humans were only internal test audiences. Old Sammy will extract various ideas from the database every day, outline the framework of the story, and then select some suitable characters and lines, use digital technology to integrate them, and then a movie will come out with a click of duang.

As for the employees of the company, they are only internal test audiences. Ai judges whether the movie is likable by monitoring the audience’s physiological feedback on the movie in real time, and whether it can mobilize the audience’s emotions and resonate. Under continuous optimization iterations, In the end, the greatest work is created.

Seeing this, I don’t know what you think of? In reality, there is a similar company, and it is Netflix . Before 2013, Netflix only distributed the films of other film and television companies to users through streaming media or mailing DVDs.

But this company is not satisfied with such a business model, and they ultimately make decisions on content creation projects by collecting user preferences. In 2013, the company analyzed nearly 30 million playback behaviors, 4 million reviews, 3 million searches and other information every day. Through the analysis of these data, they clarified the key links such as creative direction, broadcast form, and promotion methods, and finally decided to invest up to 100 million US dollars in the development of the “House of Cards” project.

---

Assuming that artificial intelligence technology will fully blossom in the future, and general artificial intelligence can be used by the public at will, what kind of interaction method is the best? computer? Or a phone or a watch? Or smart glasses?

Not at all, as one of the co-founders of OpenAi, Elon Musk has long thought about how to interact with general artificial intelligence, which is the brain-computer company Neuralink he created the following year. The company’s goal is to develop brain-computer interface technology, realize barrier-free connection and more efficient interaction methods of human-computer interaction, and help people closely integrate consciousness, thinking and digital technology to create more innovative application scenarios and business value .

This is simply developed for artificial intelligence technology. Many science fiction novels describe that the human body can be implanted with chips to obtain stronger functions or abilities. This may be possible in the not-too-distant future.

But whether ready access to information can really benefit ordinary people, I don’t think so. At the moment when short videos are popular, this amount of information is enough to make the general public unable to distinguish right from wrong, clamoring for online battles. If everyone had a brain-computer interface, wouldn’t it be Cyberpunk 2077 every day? .

postscript

It is undeniable that being able to obtain information accurately and quickly can indeed improve productivity by a large margin. Usually it takes half a day to write a copy, and it may be done in an hour. But if some people are proud, others must be frustrated. If you are not good at using tools, you will inevitably lose your job under the competition of your peers. So I think artificial intelligence technology will only benefit a few people, and the general public may not be as good as before. The monopoly of technology only puts the leading company or country above everyone.

But I am very relieved to see that ChatGPT has so many mistakes, because if there is a perfect artificial intelligence now, the life of a low-end population like me will be even more difficult to sustain.