In the field of AI, the focus has shifted from academia to industry, and the imbalance continues.
Science published an article in the latest issue of the official issue, which compiled relevant data in the field of AI in recent years. The results are also clear at a glance:
AI competition, academic circles lose .
Science counts the proportion of papers, AI largest models, and SOTA models in various fields.
Since 2016, the proportion of papers published by the industry at leading artificial intelligence conferences has started a trend of brutal growth.
In 4 years, the industry has "taken" almost 20% from the academia in terms of the number of papers.
The top 10 models in the AI field were still dominated by academia before 2013.
After that, the industry continued to exert its efforts. By 2016, almost all of the next 10 major AI models came from the industry.
Not to mention the SOTA model. Last year, the industry directly dominated the SOTA of language models and image classification,
The SOTA that analyzes emotion, semantic segmentation and target detection is roughly half the work of industry and academia, while the SOTA of machine translation all comes from academia.
Of course, this trend is also a tacit fact in the hearts of netizens, and some people even made meme pictures to ridicule:
There is a heavy brain drain in academia
The most important reason for the current situation is:
talent loss.
Taking data from North American universities as an example, computer Ph.D.s specializing in AI research are currently pouring into the industry:
In 2004, only 21% of Ph.D.s chose to enter the industry. By 2020 , the proportion of PhDs entering the industry has reached nearly 70%.
Moreover, this data is limited to talents in the field of AI research.
It can be seen from the data in recent years that the industrial demand for general computer science has not changed significantly. On the contrary, the market demand for talents specializing in the field of AI has increased eightfold since 2006.
So why are AI talents flowing from academia to industry?
From a large level, it is related to the computing power mentioned before by Science.
Obviously, the industry has a great advantage over academia in this area.
If a worker wants to do his job well, he must first sharpen his tools. Whoever has strong computing power will naturally be able to attract more talents.
But to put it bluntly, whether the computing power is strong or not depends largely on capital investment.
Give a chestnut
The source of scientific research funds in academia mainly comes from government support. In 2021, the investment by relevant US departments in artificial intelligence is 1.5 billion US dollars, and the EU's investment in the same year is 1.2 billion. Dollar.In contrast, the global AI industry’s expenditure this year has exceeded 340 billion US dollars. Even in 2019, Alphabet, Google’s parent company, invested US$1.5 billion in its subsidiary DeepMind.
Of course, the above are all discussed from a relatively macro perspective, so how do individual researchers as individual think about it?
Researcher Rowan Zellers, who just joined OpenAI after graduating from his Ph.D., showed up and gave his own account.
Rowan Zellers bluntly said that he was torn between industry and academia when choosing employment, but in the end he chose industry. And the reasons, he also listed them one by one:
- It will become increasingly difficult for academia to conduct groundbreaking system construction research
- The cost of scientific research in the field of AI is increasing exponentially
- Academia’s shift to applied research is The general trend
......
From basic research to application, the industry has it all
And the loss of talent and the tilt of computing power can also be said to be the inevitable result to a certain extent , in the Science article, two reasons are listed:
- The AI field is unique compared to other subject areas;
- The industry pays more attention to technology commercialization.
Let’s first talk about this particularity in the AI field.
In other subject areas, academia and industry will naturally form a division of labor. Basic research is completed by universities, and applied research and development are the work of industry.
However, this set of logic does not apply to the AI circle. In the field of AI, the boundaries between basic research and applied research are blurred.
In other words, there is overlap between application models used in industry and basic research. Take the Transformer model developed by Google Brain in 2017 as an example. It not only belongs to basic research, but can also be further used. directly used in industry.
In addition, the increased investment in artificial intelligence by the industry may lead to the commercialization of technology, which will not only provide substantial benefits to society, but also provide rewards for the industry itself. .
As for academia, the majority of funding sources depend on allocations from relevant institutions.
For AI talents who stay in school, although they can get bonuses and get promoted for publishing papers, scientific research is not their only job. They also have teaching tasks.
And generally speaking, academic research laboratories are non-profit. For example, Rowan Zellers worked at the Allen Institute for Artificial Intelligence before joining OpenAI: it took a lot of time and money. . .
emmmmm, it seems that the academic community has completely lost in this AI competition.
One More Thing
Of course, it cannot be said that the academic community has completely lost the AI competition. In the top 100 paper citations ranked by Zeta Alpha statistics, there is still a balance between academia and industry. Very good.
△Picture source: Zeta Alpha
Speaking of which, what do you think of the AI competition between academia and industry?
Reference link:
[1]https://www.php.cn/link/f812291e86e5d515984dc44cf9d41ac3
[2]https://www.php.cn/link/6a21bd02b3d17059c61bcb2eeb48b8cd
[3]https://www.php.cn/link/32b9e74c8f60958158eba8d1fa372971
The above is the detailed content of Science: AI competition, academia loses. For more information, please follow other related articles on the PHP Chinese website!

ai合并图层的快捷键是“Ctrl+Shift+E”,它的作用是把目前所有处在显示状态的图层合并,在隐藏状态的图层则不作变动。也可以选中要合并的图层,在菜单栏中依次点击“窗口”-“路径查找器”,点击“合并”按钮。

ai橡皮擦擦不掉东西是因为AI是矢量图软件,用橡皮擦不能擦位图的,其解决办法就是用蒙板工具以及钢笔勾好路径再建立蒙板即可实现擦掉东西。

虽然谷歌早在2020年,就在自家的数据中心上部署了当时最强的AI芯片——TPU v4。但直到今年的4月4日,谷歌才首次公布了这台AI超算的技术细节。论文地址:https://arxiv.org/abs/2304.01433相比于TPU v3,TPU v4的性能要高出2.1倍,而在整合4096个芯片之后,超算的性能更是提升了10倍。另外,谷歌还声称,自家芯片要比英伟达A100更快、更节能。与A100对打,速度快1.7倍论文中,谷歌表示,对于规模相当的系统,TPU v4可以提供比英伟达A100强1.

ai可以转成psd格式。转换方法:1、打开Adobe Illustrator软件,依次点击顶部菜单栏的“文件”-“打开”,选择所需的ai文件;2、点击右侧功能面板中的“图层”,点击三杠图标,在弹出的选项中选择“释放到图层(顺序)”;3、依次点击顶部菜单栏的“文件”-“导出”-“导出为”;4、在弹出的“导出”对话框中,将“保存类型”设置为“PSD格式”,点击“导出”即可;

ai顶部属性栏不见了的解决办法:1、开启Ai新建画布,进入绘图页面;2、在Ai顶部菜单栏中点击“窗口”;3、在系统弹出的窗口菜单页面中点击“控制”,然后开启“控制”窗口即可显示出属性栏。

Yann LeCun 这个观点的确有些大胆。 「从现在起 5 年内,没有哪个头脑正常的人会使用自回归模型。」最近,图灵奖得主 Yann LeCun 给一场辩论做了个特别的开场。而他口中的自回归,正是当前爆红的 GPT 家族模型所依赖的学习范式。当然,被 Yann LeCun 指出问题的不只是自回归模型。在他看来,当前整个的机器学习领域都面临巨大挑战。这场辩论的主题为「Do large language models need sensory grounding for meaning and u

引入密集强化学习,用 AI 验证 AI。 自动驾驶汽车 (AV) 技术的快速发展,使得我们正处于交通革命的风口浪尖,其规模是自一个世纪前汽车问世以来从未见过的。自动驾驶技术具有显着提高交通安全性、机动性和可持续性的潜力,因此引起了工业界、政府机构、专业组织和学术机构的共同关注。过去 20 年里,自动驾驶汽车的发展取得了长足的进步,尤其是随着深度学习的出现更是如此。到 2015 年,开始有公司宣布他们将在 2020 之前量产 AV。不过到目前为止,并且没有 level 4 级别的 AV 可以在市场

ai移动不了东西的解决办法:1、打开ai软件,打开空白文档;2、选择矩形工具,在文档中绘制矩形;3、点击选择工具,移动文档中的矩形;4、点击图层按钮,弹出图层面板对话框,解锁图层;5、点击选择工具,移动矩形即可。


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

Atom editor mac version download
The most popular open source editor

Dreamweaver Mac version
Visual web development tools

Safe Exam Browser
Safe Exam Browser is a secure browser environment for taking online exams securely. This software turns any computer into a secure workstation. It controls access to any utility and prevents students from using unauthorized resources.

DVWA
Damn Vulnerable Web App (DVWA) is a PHP/MySQL web application that is very vulnerable. Its main goals are to be an aid for security professionals to test their skills and tools in a legal environment, to help web developers better understand the process of securing web applications, and to help teachers/students teach/learn in a classroom environment Web application security. The goal of DVWA is to practice some of the most common web vulnerabilities through a simple and straightforward interface, with varying degrees of difficulty. Please note that this software

mPDF
mPDF is a PHP library that can generate PDF files from UTF-8 encoded HTML. The original author, Ian Back, wrote mPDF to output PDF files "on the fly" from his website and handle different languages. It is slower than original scripts like HTML2FPDF and produces larger files when using Unicode fonts, but supports CSS styles etc. and has a lot of enhancements. Supports almost all languages, including RTL (Arabic and Hebrew) and CJK (Chinese, Japanese and Korean). Supports nested block-level elements (such as P, DIV),