AI is developing rapidly, which industry jobs may be replaced?

AI 0 529

The first wave of AI unemployment arrived on International Labor Day, May 1 this year, when tech giant IBM announced it was suspending hiring for 7,800 people, saying that jobs in those positions would be replaced by AI.


Previously, at the end of March, Goldman Sachs Group released a report that predicted 300 million jobs worldwide would be replaced by generative AI, with lawyers and executives being the most affected. Countries around the world are also seeing the first batch of unemployed designers and copy editors one after another because of ChatGPT and Midjourney.


In the next 3-5 years, what kind of jobs will be replaced by AI? Which industries are relatively safe? What kind of skills are needed if you want to become an AI engineer? And can liberal arts students switch to AI?


The emergence of ChatGPT-4 is shocking, AI will have such a thing out sooner or later, but I didn't expect it to be so fast and work so well. Many people are currently taking screenshots of ChatGPT-4's chats and are very fanatical. Together with Midjourney V5, everyone is worried if their jobs will be replaced by AI.


mj.jpg


This concern is justified. The most amazing thing about ChatGPT-4 is its "emergent function", that is, when it trains a large enough amount of data, this complex system is born with properties that its components do not have - close to human "thought patterns" and "intellectual performance".


There is a chain of thought inside that helps ChatGPT-4 to "chain thinking". Just like we sometimes do homework, at a certain point, we can't do it, and then the parents say "think again", which actually doesn't say anything, but the student thinks I may still have some things to master, and through slow thinking and a little guidance, he suddenly gets a correct answer.


So you ask ChatGPT-4 to "think again" in the dialog box, and it will also give you an improved answer again, and people will feel surprised.


Because of AI's increase in productivity, the phenomenon that one good person can do a lot of work and run a large-cap company by a small group of people may become more and more common in the future. You see Midjourney is a typical example, with only 11 employees, but annual revenue of $100 million.


So which industries may be more affected?

1、Translation and customer service

Translators and customer service have been replaced almost, now answer the phone courier, many robots. The company's main goal is to provide the best possible service to its customers.

ChatGPT is embedded in Office365 and can automatically generate presentations and tables

The next most dangerous ones are office clerks, human resources, and those who do financial statements. Microsoft Office365 has embedded ChatGPT into Word, PPT and Excel, which can automatically generate presentations, PPTs and tables, so the value of these Office skills that you used to learn so hard has gone down.

There is a paragraph that says "finance will not be replaced by AI, because it can not do scapegoat", although a little bit of truth, but the increase in productivity means that the company's demand for financial talent compression, your employment space becomes smaller.


2, the lawyer industry

There is also the lawyer industry. We know that an important part of a lawyer's job is to be proficient in law and find past cases, and the process of finding them is very time-consuming, so there should be a part of the law firm dedicated to this work.

The company's main goal is to provide the best possible service to its customers.


3. Programmers may also be affected

ChatGPT-4 also generates code's, and some programmers will be affected, especially on the front end. The programmer will be able to use the software to create a web page.

From the company's point of view, it is possible that ChatGPT will be more inclined to write code in the future. Because everyone's style of writing code is different, an employee leaves and a new employee comes in and may have to rewrite the code because it doesn't work well. Then ChatGPT's consistency will be better and more efficient from the company's point of view.


4. Illustrators and designers may be affected

Illustrators and designers affected by Midjourney, I read online that some of them have been laid off. It may take you 2 days to finish an illustration by a human, the machine comes out in minutes and the result is still very good, which is forcing people to do more innovative work.

It is said that now some universities in the US, in the direction of natural language processing, computer vision and speech recognition, no longer increase the number of faculty positions.

Then we discussed why? In the past, research institutions, 3-5 years will produce some results, fine and fragmented need so many people to do, but ChatGPT-4 came out, it has solved many problems, the rest are some very difficult hard bones, then you are not need so many faculty to do, it led to some positions were reduced.


Which industries are not easily replaced by AI artificial intelligence?

First of all, jobs related to entities, such as doctors, nurses, drivers, and niche craftsmen, such as those who make guqin and artists who make ceramics, are dependent on personal experience to do, and are less likely to be replaced by AI.


Because AI has been doing mostly cognitive-related tasks, perception of this piece of work less, at this stage and the entity related to do not do well, compared with humans, robotic hands are relatively elementary, screwing a bottle cap is still a difficult thing.


Even cleaning, for us humans is "simple and easy to repeat", but for the machine is a vague concept, there is no way to program or formalize.


The company's main goal is to provide a better solution to the problem.


We think about how ChatGPT is up? It's data are Billion, that is, more than 1 billion, which means so much data, most likely are not set up privacy in order to be called by it.


If an industry involves privacy, data can not be made public, can not be on the model training, then AI can not squeeze in. Areas like healthcare, banking, and biology, for example, are relatively safe.


If high school students choose a major and only consider the employment prospects, I think the direction of artificial intelligence is still the best, the so-called "not into the tiger's cave how to get a tiger's son.


The new term "AI for Science" is used to help the development of science, and in the future, all industries will need the assistance of AI, which will be operated by people who know the direction of AI, so there will be a very large talent gap.


A good AI researcher or engineer needs three basic qualities: mathematical foundation, programming ability, and English. Learn English because you need to follow the most cutting-edge international technology and read literature, and then the requirement for programming ability is a bit higher than math.


It is not like before that you need to know particularly deep knowledge of artificial intelligence, and if you are a computer or other science major, the threshold is not so high if you switch to AI.


First of all, most of the research now is modular, deep networks are some models that are being built like building blocks. In terms of algorithms, on ArXiv you are able to quickly know what the latest algorithms are like, and the code itself is shared on many sites, such as Github. These three points make it relatively easy to enter the industry now.


The current mainstream development path of AI is three major blocks: models, computing power, and big data.


The pessimistic part is that the model framework predecessors have done well, almost publicly, and researchers make it bigger and deeper on the line.


The first time I saw a new deep learning model, Geoffrey Hinton came up with a deep learning model in 2006, and then there was an image classification competition that used the large-scale dataset ImageNet. percentage points over the previous winner.


What is this concept? If you were doing it with traditional machine learning methods, it would improve by 0.3 to 0.4 percentage points per year. That means that the deep learning approach is about 20 years faster than the traditional machine learning approach. So at that time, everyone turned to do deep learning models.


But deep learning models require powerful computing power to run on top of a specific GPU chip. It is said that ChatGPT has 10,000 A100 GPUs to support, and the price of a single A100 is about $10,000, so the cost of GPU alone is $100 million (about 600 million RMB), which is one of the reasons why OpenAI is a small company with less than 100 people, and Microsoft has invested $2 billion to go up. So big models can only be done by big companies and organizations.


Historically speaking, artificial intelligence is less than 90 years old, and we generally consider its beginning to be the Turing machine in 1936, during which it has been going through a process of ups and downs.


It went through its first winter in the 70s and 80s, when you couldn't get a project if you said you were in artificial intelligence. In the early 90s, it went through a second winter.


It was in 2012, the year Geoffrey Hinton won the competition with his students, that AI really took off.


By 2016 AlphaGo won Lee Sedol, and then in 2017 Google researched the Transformer network, which was followed by a series of ChatGPT efforts, as well as self-driving, AI finance, AI healthcare and various other fields moving forward.


But in fact, by 2022, the AI industry is a bit of a downward trend, because we feel that what should be done are done, and do not see a good application, it is clear that some large companies of deep learning this section, has been laying off staff. But suddenly in March this year all of a sudden ChatGPT-4 came out, it pulled everyone back again.


So it has a boom and a bust period. I've been in this field for a long time, so I think I'm a bit more calm about the ChatGPT-4 boom. the scope of AI research is very broad, many problems are difficult to realize in a short time, and there is still a long way to go for human understanding of intelligence.


May be interested in

Comments0

Comment Now

Feel free to express your views.