Behind AI unemployment anxiety: more than 10 million people have found new jobs!

Original source: Huashang Taolue

Image source: Generated by Unbounded AI‌

Hollywood, USA, 160,000 people announced a collective strike to protest AI invasion…

But at the same time, on the Loess Plateau thousands of miles away, many women who did not go to college found jobs because of AI and changed their lives.

They are feeding the AI

Qingjian, Shaanxi Province, is located in the Loess Plateau.

In the past, the only specialty here was red dates. However, in recent years, with climate change, jujube production has decreased, and many young people have left their hometowns to make a living in other places.

The 30-year-old Cao Yali is one of the rare retrogrades.

In 2019, due to family changes, she returned to her hometown in Qingjian from Xi’an. But it is difficult to find a job in a small county. Cao Yali has been at home for more than a year and is very anxious.

Until 2020, a company called Qingjian Aidou changed her fate.

The reporter of “Nanfengchuang” magazine recorded Cao Yali’s story through interviews.

In Qingjian Idol, Cao Yali became a national AI data labeler. **

The daily work of her and her colleagues is to mark text, images, videos, etc. through picture frames, labeling, sorting, and finding differences.

It is because of their hard work that feeds a large amount of data to AI, and the AI model is getting smarter and smarter.

Although this job is not easy, an average of 30,000 pictures are viewed every day. But Cao Yali was still very excited when she thought of saying goodbye to the life of facing the loess and turning her back to the sky.

“The mouse is money at all.”

Compared with Cao Yali, the 34-year-old Wang Hui chose this job more because she was tired of being a full-time mother.

Three years ago, she was still a full-time mother, devoting herself to her children. Although this kind of life is not bad, it always feels like something is missing.

By chance, Wang Hui saw the recruitment advertisement of Qingjian Idol in Moments, and was moved.

“What do data labelers do? I’ve never heard of it before.” For Wang Hui, who has been at home for many years, whether she can still go out to work is a question mark, let alone such a trendy job.

But after training and her own hard work, she quickly got started.

“I used to feel that it was difficult to change my life, but now I understand that as long as I seize the opportunity, there are many possibilities in the future.”

Not only Cao Yali and Wang Hui, but also Qingjian in Shaanxi.

In recent years, with the explosion of AI, the demand for data labelers has surged. In Shanxi, Guizhou… This new profession has also been added in many parts of the country, and there are many practitioners.

Among the major Internet companies, ByteDance alone has as many as 40,000 data labelers in Jinan, Tianjin, and Wuhan.

If you add third-party data labeling service companies, such as the country’s largest Testin cloud measurement, and the aforementioned Qingjian Idol, the total number of practitioners is even more amazing.

According to the “Science and Technology Daily” report, in Beijing alone, there are more than 100 companies engaged in data labeling, and there may be more than 10 million people engaged in this work across the country. **

AI has not only changed the fate of many people, but also changed the economic landscape of the Midwest.

Qingjian, Shaanxi is a famous old revolutionary base area.

However, limited by the geographical environment, the economy here has not been ideal. During this period, I have tried many industries such as raising pigs, planting fruit trees, and animal husbandry.

“Everything that people can think of in rural areas has been done in our county, and almost all of them have failed.”

In contrast, AI data annotation has become a rare industry that has successfully gained a foothold here. Today, Qingjian Aidou has become the largest employment enterprise in **. **

Fengjie, Chongqing, is the location of the Baidi City where “the court resigned to Baidi among the colorful clouds, and the thousand miles of rivers and tombs returned in one day”. It is also the gateway to the Three Gorges where thousands of poets such as Li Bai, Du Fu, Bai Juyi, Su Shi, and Su Zhe lived or served as officials and left tens of thousands of poems.

There are picturesque magnificence and beautiful mountains and rivers here, but there has been a long-term absence of modern industrial enterprises, and it has been wearing the hat of a poor county for many years.

In a place where there are not many factories, it was unthinkable in the past to introduce companies on the top of the wave such as Baidu and JD.com. However, the leaders of the new county party committee and county government, with keen eyesight, bold imagination, and careful verification, decided to find a place in the digital economy wave, and ushered in digital enterprises such as Baidu, JD.com, and Netease, and integrated into the information industry and the digital economy era.

Among them, the Baidu Smart Cloud Data Industry Base is also staging the story of Baoma becoming a data labeler, and has grown to a scale of more than 500 people in just one year, becoming a new growth pole for local employment.

Behind the AI is a lot of manual work

Artificial intelligence, in the subconscious of many people, means that intelligence replaces artificial intelligence and a large number of unemployment.

But what is little known is that AI is inseparable from the feeding of a large number of unsung heroes, and they are ** data taggers. **

Why does AI need data feeding? This touches on the difference between machine learning and humans.

We know that people have the ability to infer other things from one instance. For example, when we see a cat, we can recognize it as a cat, or even what kind of cat it is.

But the machine is different. It needs to draw inferences from others and learn how to recognize a cat through a large amount of data.

So, who will feed AI data? The answer is data labelers. In fact, the higher the quality of the data fed, the more powerful the AI.

Therefore, there is a saying in the industry: the intelligence is only as powerful as artificial intelligence. **

In 2005, when artificial intelligence was still in its infancy, industry leader Zhu Chunsong rushed back from the United States to his hometown of Ezhou, Hubei, where he founded the Lianhuashan Research Institute.

It is said that there was the earliest big data labeling team in the world at that time.

But at that time, data quality had not received widespread attention, and artificial intelligence continued to develop slowly along the path of algorithms and models. Until 2012, an AI model called AlexNet became a blockbuster in the ImageNet challenge.

ImageNet is a large database with 14 million labeled images.

Famous Chinese computer scientist Fei-Fei Li is one of the initiators of the database project. Before that, the common research direction in the industry was models and algorithms.

Li Feifei took another approach, starting with data quality.

Today, ImageNet is already the world’s largest image recognition database, which gave birth to a new wave of artificial intelligence in 2012.

Behind this database, there are nearly 50,000 data labelers from more than 160 countries. It took them three years to label all these pictures.

Without these annotated images, AI’s historic leap would not have been possible.

How important is data labeling? Perhaps it can be illustrated with a story circulating in the industry.

A few years ago, in the United States, a self-driving car of a certain brand got under a large truck, which eventually led to the tragedy of car crash.

The investigation of the accident revealed that the car’s camera system mistook the white truck for Baiyun, and the reason was: the ** data labeling was not done properly! **

“We are like the eyes of AI, take it to see the world, understand the world, and grow rapidly.” A practitioner described the importance of data labeling in this way.

It is thousands of data labelers who draw boxes day after day and teach artificial intelligence to understand the human world.

However, it is ChatGPT that really makes data labeling popular.

In 2019, OpenAI founder Sam Altman proposed a bold idea to Microsoft: ** Build an AI system that will forever change the way humans interact with computers. **

This is the ChatGPT that is popular all over the world today.

In order to build ChatGPT, OpenAI not only needs a large investment from Microsoft to purchase computing power, but also needs to solve a problem:

**Let the big model learn to recognize violence, hatred and other offensive speech. **

This process requires a lot of manpower to label offensive speech. As a result, OpenAI found an outsourcing company Sama to reach a cooperation.

The data is provided by OpenAI, and then Sama hires people in Kenya, Africa, to label the data.

Although Sama’s behavior has been criticized by many media as a sweatshop, data labeling has brought a lot of job opportunities to the local area.

Today, under the background of the explosion of artificial intelligence, **Sama has become the secret weapon of many AI giants around the world, including Google, Microsoft and OpenAI. **

This process, also on a global scale, has created a large number of new jobs.

Don’t be a Luddite

In human history, almost every round of technological revolution has been accompanied by people’s anxiety about unemployment.

In the 19th century, as machines were widely used in British factories, a large number of skilled workers were forced to lay off. Among them, a man named Lude, in a rage, led the workers to smash the machine.

These people later became the object of ridicule, and economists nicknamed them: Luddites.

The Frenchman Bastiat taunted ruthlessly:

“You say that new technology destroys employment, so can technological regression increase wealth?”

After the rhetorical question, Bastiat gave an example. If you find that the new big ax is too easy to use, just replace all the big axes with small axes, so that the tree that can be cut down in one day will take three days to cut down.

Will this create more jobs and support more loggers?

Bastiat’s taunt was a fatal blow to the Luddites.

Afterwards, the development of human society has further proved that the anxiety of the Communist Luddites is actually unfounded. **

The popularity of machines has not eliminated human employment. On the contrary, technological progress has made human division of labor more detailed, production efficiency is getting higher and higher, and more labor is required.

The latest example is that many people complain that Internet e-commerce has overwhelmed physical stores.

But at the same time, it has also spawned many new jobs such as couriers, takeaway boys, online car drivers, programmers, etc., which were unimaginable in the past.

In fact, in China today, there are as many as 84 million workers in new forms of employment, such as couriers and food delivery workers.

AI is no exception.

While it is displacing some traditional jobs, it is also creating a large number of new jobs. Data labelers are just the tip of the iceberg.

In the future, the society will inevitably generate a large demand for positions such as image recognition, algorithm research, and deep learning. These are all new employment opportunities generated by artificial intelligence.

This does not include the thousands of industries that will be completely restructured because of AI.

In this regard, Goldman Sachs Group has published a research report “AI’s Potential Huge Impact on Economic Growth”. In this report, Goldman Sachs came up to express its position and views:

**Judging from historical experience, the jobs replaced by AI will be offset by the new jobs it creates! **

However, that doesn’t mean we can sit back and relax. In fact, since the ChatGPT fire, it has indeed brought employment crisis to many people.

For example, Hollywood actors, screenwriters, and photographers can’t sit still.

160,000 people declared a collective strike, which is unique in the history of Hollywood and is destined to be written into film history.

But strikes cannot stop mankind from entering the AI era. In this regard, the correct approach is not to smash the machine like the Luddites, but to:

**Take the initiative to change yourself to adapt to the new era! **

Reference materials:

[1] “The Artificial Power Behind Artificial Intelligence: Over 10 Million People Engaged in Data Labeling” Science and Technology Daily

[2] “On the Loess Plateau, women who haven’t gone to college are feeding AI” Nanfengchuang

[3] “Chinese County Workers Training Artificial Intelligence”, Blue Word Project