2024-08-21
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
TencentTechnology writer Li Anqi
Edited by Su Yang
21 months after ChatGPT was released,AIThe war for talent is intensifying.
BeforeOpenAILianchuang left one after another and switched to competitors.GoogleCharacter.AI’s talent and technology were acquired for US$2.5 billion. Domestic giants are also scrambling to snatch talent.
On May 20, ByteDance's Top Seed Talent Program was officially launched
On June 19, Tencent Qingyun Project was officially launched
July 23,HuaweiRecruiting talented young people from all over the world
On July 26, Alibaba launched the T-Star program to discover high-potential students
……
Whether it is the seed plan or the talented teenagers, theseThe recruitment plans of large companies all point to one key word:AI。
According to a report released by British consulting firm Dealroom in May this year, global investment in artificial intelligence will reach $65 billion in 2024, accounting for 20% of the total global venture capital. The investment in AI by the seven largest US stock companies in the first half of the year alone has exceeded the total amount of venture capital in the UK in a year.
Where there is the most money, there are also the smartest minds.
Nowadays, competition in the field of AI is not only about who has the most GPUs,HashrateThe strongest and more importantly, what kind of brains should be matched behind hundreds of thousands of GPUs to be the first to capture the flashes of inspiration and the arrival of the singularity moments behind hundreds of millions of calculations.
Silicon Valley's technology giants have just experienced an unprecedented storm of layoffs.
Over the past few years,Meta, Amazon, Microsoft, Google and Apple have cut hundreds of thousands of jobs. According to layoffs.fyi, a layoff data website, the number of layoffs in the technology industry in 2023 exceeded 260,000.
However, even though the tech giants were still laying off employees at the beginning of the year, a "talent war" was also taking place at the same time. The AI field they are targeting seems to be "the best place to be".
According to CompTIA, there will be 180,000 AI-related jobs in the United States in 2023. This year, job openings in the field of AI are still increasing.
In 2023, 27% and 18% of the positions posted by Alphabet and Meta will be related to AI, respectively, up from 23% and 8% in 2019.
The tech giants' extreme thirst for AI talent has reached an unprecedented level. Even this round of layoffs is partly due to the need to attract and retain top AI talent. On the one hand, they need to increase the salaries of core R&D personnel, and on the other hand, they need to invest more money and buy more cards.
Google CEO Pichai has made it clear that the company hopes to increase investment in cutting-edge fields such as AI and cut costs in other areas. Reducing costs and increasing efficiency is to reallocate resources and concentrate money on AI.
High salaries are the primary olive branch that giants offer to AI talents. Annual salaries of millions of dollars clearly show how much companies value AI talents.
Take AI engineers as an example. According to a survey released by Levels.fyi in May this year, the average salary of entry-level AI engineers in the United States is $239,000. Top companies like OpenAI offer annual salaries of more than $900,000.
The salaries of AI engineers at different levels are generally higher than those of non-AI engineers. This shows that large companies are willing to pay a premium for AI talents. As Vasant Dhar, a professor of information systems at New York University, said: "Artificial intelligence is the new Wall Street... Really good talents are in short supply."
For those talents that have already been recruited, companies try to retain them by increasing their salaries.。
According to data from a WTW survey of more than 1,500 employers, base salary increases for management employees in the AI and machine learning industries will range from 5% to 11% between April 2022 and April 2023, and for non-management employees will increase by 13% to 19%.
The reason why the giants are willing to raise salaries is mainly because they are afraid of being poached. In April this year, Tesla's visual scientist Ethan Knight left and joinedMuskIt sounds a bit strange that Musk is poaching people from Tesla for xAI.
But Musk later revealed that if he had not intervened, the scientist would have been poached by OpenAI.
Musk said with emotion: This is the craziest talent war I have ever seen. In addition to being generous in terms of salary, technology giants also give AI talents enough face.
It has become a new trend for bosses and executives to personally rescue people.First, Google founder Sergey Brin personally called to retain employees who wanted to change jobs; then Meta boss Zuckerberg wrote a handwritten letter, sincerely inviting DeepMind researchers to change jobs and do HR work himself.
in addition,Giants are also keen to open AI labs, research institutes or cooperate with schools around the world to reserve outstanding scientific research talents through multiple channels., discover talented young people with high potential early.
The Washington Post once reported that Silicon Valley is "killing" the academic circle. Many AI PhDs enter large companies as soon as they graduate because they provide excellent scientific research environments and resources. According to a report published in Science magazine in 2023, nearly 70% of talents with AI PhDs eventually enter private companies, while 20 years ago this proportion was only 21%.
In recent years, "artificial intelligence" has frequently appeared in government reports and strategic plans of various countries. It should be said that attracting and retaining talent has become one of the highest priorities for countries to develop AI. This is true for the United States and China across the ocean.
In the past five years, China and the United States have become the main sources and target workplaces for top AI talents.A report released by the US MacroPolo think tank pointed out that 70% of the top AI talents work in institutions in China or the United States, and 65% of the top AI talents come from China and the United States.
Despite this, the United States remains the preferred destination for the best talent.More than half of the top 2% of talent in the AI field choose to work in the United States.As China enters the first echelon of this wave of AI, the country's demand for AI talents has reached an unprecedented level.
To what extent is there a shortage of talent in the domestic AI industry?
According to a report released by McKinsey, it is estimated that by 2030, domestic and foreign universities and existing top talent reserves will only be able to provide about 2 million AI talents, and the gap will reach 4 million.
Despite this, thanks to the rapid development of China's AI industry in recent years, more and more young people are willing to stay in China for development. The MacroPolo report shows that among the top 20% of AI talents in the world, 28% choose to work in China. In 2019, this figure was only 11%.
China has cultivated the largest number of AI talents in the world, and the attractiveness of domestic companies to AI talents has also increased significantly. In order to keep top talents in China, domestic technology companies are working at full capacity.
Their determination to retain AI talents can be seen from the special AI talent plans launched by major companies in recent years.
This year marks the second year that Tencent has officially launched the Qingyun Project.Large ModelThe first year of full-scale outbreak. As a special AI model recruitment program opened by Tencent for top technical talents, the selected candidates of Qingyun Program enjoy top resource allocation.
Fu Zhiyuan (pseudonym), a doctoral student at Nanyang Technological University who just graduated in June this year, was one of the first batch of campus recruits of Tencent Qingyun Program. He has now joined Tencent IEG Photonic Technology Development Department and is engaged in training game AI.
Compared with pure academic research in an ivory tower, Fu Zhiyuan described working at Tencent as "a feeling of using AI to its full potential." As a leading domestic company, Tencent is capable of providing top-notch computing resources and gaming environment resources.
“When we train the model and run it in a real business environment, we feel a strong sense of accomplishment, just like hearing the roar of a racing car engine. We feel excited and looking forward to it.”
In addition to ample financial computing power and industry resources, in order to attract the best talent, large companies also provide fresh graduates with double insurance of "salary" + "development prospects".
According to the 2024 College Graduate Employment Data Report released by Liepin, salary and benefits are the primary factors that graduates consider when choosing a job. Compared with the survey results of the same period last year, it was found that graduates have become more concerned about corporate prospects and personal development.
A fresh graduate who joined the AI department of a leading company revealed to Tencent Technology,Currently, the salaries offered to fresh graduates with a master's degree in large model-related positions in the market are basically over 500,000 yuan. If they are doctoral students, the salary will be even higher.This information is also confirmed in the "2023 Artificial Intelligence Talent Insight Report" released by Maimai Gaopin Talent Think Tank: Starting from June 2023, the average monthly salary of newly created AIGC positions will stabilize at around 60,000.
As attractive as the high salary is the complete training mechanism promised by the big companies. This usually appears in the recruitment notices of the top companies with words such as "professional mentor", "targeted growth" and "exclusive training".As the overall environment becomes more uncertain, new graduates are beginning to value the long-term value that their work can bring to themselves.
"Since the company dares to let me create something from 0 to 1 that no one has ever done before, what should I be afraid of as a graduate?" After graduating with a Ph.D. from Tsinghua University, Kong Tao joined ByteDance through the talent program and engaged in the research and development of AI robots.
After joining the company, he immediately initiated a project and focused on robot research. Companies usually give special recruits enough trust and patience. According to Kong Tao, everything from project establishment to approval was very smooth. Now he is the director of robot research at ByteDance.
In addition, the diversified business formats and abundant cash flow of large companies also provide a place for academics who only want to do research. Wang Aiwen (pseudonym), who graduated with a doctorate and joined Tencent AI Lab, said that she enjoyed a very pure research environment in Tencent AI Lab. Since this year, three studies she participated in have been published in top international academic journals.
In addition to having complete hardware, each major company also has its own unique "soft technology" when competing for talent.
According to an HR in charge of campus recruitment at a large Internet company,In the past two years, in order to "pick the best" in the autumn recruitment, everyone will "get ahead" in advance. Opening offline interview sessions and booking core laboratories for prospective doctoral students are all basic operations.。
So far, Tencent’s Qingyun Project has held nearly 10 offline events in Beijing, Shanghai, the Bay Area of the United States, and other places, inviting top technical students to visit and exchange ideas.
It is understood that in order to increase its appeal, during the ACL conference (Annual Meeting of the Association for Computational Linguistics), the Qingyun Project even held the "Technical Exchange Dinner" directly at the venue of the conference - a cruise ship on the Mekong River.
"The 'early batch' is locked in July and August, and letters of intent are sent out in September and October. The top graduates that everyone is eyeing are basically 'offer harvesters'. If you act a little later, they will be poached away."
What’s interesting is that the venues of this year’s top academic conferences have become large-scale recruitment sites for domestic companies to poach talent.
According to participants who attended CVPR (IEEE International Conference on Computer Vision and Pattern Recognition, one of the three top conferences in the field of computer vision) this year, Tencent, as one of the sponsors of this year's conference, not only had relevant core technical leaders attend the discussion, but also directly sent HR to the site.
There are also reports that at this year's ICML (International Conference on Machine Learning, one of the three top conferences in the field of machine learning), the business and HR of the ByteDance team came to the scene to check the number of papers published in Google Scholar, and directly interviewed and issued offers to suitable candidates.
At the time of the palace fighting scandal at the end of last year, OpenAI officially stated: Without employees, OpenAI is nothing.
After all, OpenAI's core technological breakthroughs almost all come from the authors and research teams of related papers, and talent is decisive for the development of AI technology.
For domestic large companies, whoever has more talent reserves is closest to China's OpenAI. Internet companies, which have always pursued ROI, spend so much energy and effort to snatch talent out of a survival instinct.
Regarding talents, Ren Zhengfei once told such a story - Huawei once recruited a young Russian PhD in its early years. He was not good at dating and could only do mathematics. Even his supervisor could not explain what he had been tinkering with every day for more than ten years.
But suddenly one day, this Russian guy announced that he had achieved a technological breakthrough through an algorithm, making a base station compatible with both 2G and 3G networks. This allowed Huawei to suddenly surpass Ericsson and occupy the European market on a large scale.
This kind of "technological explosion" also applies today.It seems that big companies are competing for talent, but in fact they are afraid of missing out on the ticket to a new round of opportunities.These people dubbed as "genius teenagers" naturally take on the anxious expectations of the big companies.
Regarding AI competition, there is a generally recognized rule in the industry: talent first, data second, and computing power third.
Huang Wenhao, head of model training at Zero One Everything, also pointed out that the role of talent in large models will be magnified, and the upper limit of a company = talent quality × computing power. The large model algorithm team is generally ten to twenty people, but they will use tens of thousands of cards later, and the capabilities of these people are greatly magnified by the tens of thousands of cards behind them.
Moreover, due to various reasons, domestic large companies are far from being as generous as OpenAI in having tens of thousands of cards to use. The only thing they can grab is this group of talented young people who have ideals for technology and are eager to show their talents.
References
https://cn.nytimes.com/technology/20240207/why-is-big-tech-still-cutting-jobs/
https://m.yicai.com/news/101960767.html
https://www.levels.fyi/blog/ai-engineer-compensation-q1-2024.html
https://36kr.com/p/2687911763979651
https://www.mckinsey.com.cn/%e5%9c%a8%e5%8d%8e%e4%bc%81%e4%b8%9a%e5%a6%82%e4%bd%95%e5%a1%ab%e8%a1%a5ai%e4%ba%ba%e6%89%8d%e7%bc%ba%e5%8f%a3/
https://macropolo.org/digital-projects/the-global-ai-talent-tracker/
Collect and organize the latest news information and make it available for free for everyone to read
