Beating the table with the boss, power being lifted, income not improving: scientists are "escaping" from the enterprise

人才的故事总是离不开行业的发展背景。

The story of talents is always inseparable from the development background of the industry.

我们需要回到2016年AI浪潮的开始。

We need to go back to the beginning of AI wave in 2016.

当时，人工智能风起云涌，中国的创业热情空前高涨。一些人工智能明星公司应运而生。他们募集的资金往往超过1亿元，估值一路走高。例如，尚唐在2017年至2018年间筹集了数十亿美元。仅2018年，融资总额就达到22.2亿美元，估值从2017年底的20亿美元飙升至60亿美元。

At that time, the wind of AI was just rising, and the enthusiasm for entrepreneurship in China was unprecedented. A number of AI star companies emerged. They often raised more than 100 million yuan, and the valuation was rising all the way. For example, Shangtang raised billions of dollars between 2017 and 2018. In 2018 alone, the total amount of financing reached $2.22 billion, and the valuation soared from $2 billion at the end of 2017 to $6 billion.

公开资料显示，2016年至2018年，人工智能行业共发生投融资事件2000余起，融资金额再创新高，达2500亿元。资本的流动往往决定了顶尖人才的流动。

According to public records, from 2016 to 2018, there were more than 2000 investment and financing events in the AI industry, and the amount of financing reached a new high of 250 billion yuan. The flow of capital often determines the flow of top talents.

这种背景给多年坐板凳的大学科学家带来了巨大的机遇。

This kind of background has brought great opportunities for University scientists who have been on the bench for many years.

一位知名的AI牛曾在接受老虎嗅觉采访时说，"AI兴起之前，我们还处在研究生和博士阶段。我们不好意思告诉别人我们在做算法，但说我们在学计算机，因为当时很多企业不懂算法和人工智能。相反，电脑更受欢迎。"

A well-known AI bull once said in an interview with tiger olfactory, "before the rise of AI, we were in the graduate and doctoral stage. We were embarrassed to tell others that we were doing algorithms, but said that we were learning computers, because many enterprises didn't understand algorithms and AI at that time. On the contrary, computers were more popular."

当人工智能流行时，这些以前的研究人员成为企业竞争的对象。为了争夺这些人才，许多公司甚至发起了竞价战。

When AI became popular, these former researchers became the object of competition for enterprises. Many companies even launched bidding wars in order to compete for these talents.

卡内基梅隆大学计算机科学学院院长安德鲁·摩尔教授曾公开表示："一位人工智能专家对一家企业的价值至少是500万至1000万美元。"。

Professor Andrew Moore, Dean of the school of computer science at Carnegie Mellon University, once said publicly: "the value of an AI expert to an enterprise is at least US $5-10 million.".

据知情人士透露，对于人工智能领域的知名专家来说，他们的年薪和股票总收入在四五年内将达到几百万或上千万元。在未来的某个时候，他们会像职业运动员一样续约或重新谈判新合同。

According to people familiar with the situation, for well-known experts in the field of AI, their annual salary and stock total salary will reach several million or tens of millions of dollars in four or five years. At some point in the future, they will renew or renegotiate a new contract, much like professional athletes.

例如，据新浪科技报道，2019年9月18日，阿里宣布高通公司前总工程师陈颖和加拿大西蒙弗雷泽大学终身副教授谭平加入阿里人工智能实验室，两位科学家的年薪高达一百万美元。

For example, according to sina science and technology, on September 18, 2019, Ali announced that Chen Ying, former chief engineer of Qualcomm, and Tan Ping, lifelong associate professor of Simon Fraser University (SFU) in Canada, had joined Ali Artificial Intelligence Laboratory, and the annual salaries of the two scientists were as high as one million US dollars.

一方面，当商业模式不明确时，他们需要科学家的平台。在那个时候，似乎"那些得到人工智能的科学家得到了一切"。另一方面，由于深度学习等技术太新，人才积累不足。稀缺性使得人工智能人才的流动越来越频繁。科技巨头们千方百计抢夺顶尖人工智能人才，如百度的"青年指挥员计划"、阿里巴巴的达摩学院、Facebook的fair等。

On the one hand, when the business model is not clear, they need the platform of scientists. At that time, it seems that "those who get Ai scientists get everything.". On the other hand, because deep learning and other technologies are too new, talent accumulation is insufficient. The scarcity makes the flow of AI talents more and more frequent. Technology giants try their best to grab top AI talents, such as Baidu's "young commander plan", Alibaba's Dharma academy, Facebook's fair, etc.

以阿里达莫学院为例。它成立前后，疯狂扩充人员，曾被称为顶尖科学家的"乌托邦"。包括但不限于以下专家学者：

Take the aridamo academy as an example. Before and after its establishment, it expanded its personnel crazily, and was once known as the "Utopia" of top scientists. Including but not limited to the following experts and scholars:

从上图可以看出，2019年这一悬崖峭壁般的缺口尤为明显。当人工智能企业无法再通过PPT获得融资，当业界质疑声大于肯定声时，那些最具明星光环的科学家们将率先听到失败的声音。

As can be seen from the above figure, this cliff like gap is particularly obvious in 2019. When AI companies can no longer obtain financing by PPT, when the voice of doubt in the industry is greater than affirmation, those scientists with the most star Aura will be the first to hear the voice of failure.

因此，无论是资金充裕的大型互联网公司，还是业内的明星人工智能公司，都在逐步回归理性，对人工智能科学家的"性价比"进行评估。

As a result, both big Internet companies with abundant funds and star AI companies in the industry are gradually returning to rationality and evaluating the "cost performance" of AI scientists.

对于初创企业来说，"性价比"更为重要。比如，对于一些企业来说，邀请科学家加盟，可以使企业拥有更多的品牌价值，从而获得资本市场的认可，便于融资，使企业在短时间内提高在行业内的知名度。坦率地说，科学家是企业的门面。

For start-ups, the "cost performance" is more important. For example, for some enterprises, inviting scientists to join can make the company have more brand value, thus gaining recognition in the capital market, facilitating financing, and enabling the enterprise to improve its popularity in the industry in a short period of time. Frankly speaking, scientists act as the facade of enterprises.

因此，我们可以看到，一开始，企业会拿出非常好的条件和前景来吸引科学家，但毕竟中小企业"耐心"不够，无法实现技术与产业的结合，探索周期较长。对于关注科研成果的科学家来说，这是不可接受的"妥协"。

Therefore, as we can see, at the beginning, enterprises will come up with very good conditions and prospects to attract scientists, but after all, small and medium-sized enterprises are not "patient" enough to achieve the combination of technology and industry with a longer exploration cycle. This is an unacceptable "compromise" for scientists who pay attention to scientific research achievements.

因此，人才流动的速度变得更快。

As a result, the speed of talent flow has become faster.

至于原因，有分析人士表示，实验室的运行模式可能有问题。毕竟，马云曾在2017年的云栖大会上定位实验室（包括达摩研究所）："90%以上的研究项目不仅可以在实验室，而且必须在市场。只有这样，实验室才能走得更远。"

As for the reason, some analysts said that there might be a problem with the operation mode of the laboratory. After all, Ma Yun once positioned laboratories (including Dharma Institute) at the cloud habitat conference in 2017: "more than 90% of the research items can not only be in the laboratory, but must be in the market. Only in this way can the laboratory go far. "

不过，针对这一事件，阿里巴巴方面表示，"Ai实验室在上一轮架构变革中已经整合到云智能中，目前仍在招募大量人员。"

However, in response to this incident, Alibaba said that "Ai labs has been integrated into cloud intelligence in the last round of architecture changes, and is still recruiting a large number of people."

有了这样的教训，越来越多的大工厂在引进科学家方面变得更加谨慎和聪明。

With such a lesson, more and more large factories are becoming more cautious and clever in introducing scientists.

近日，一家互联网工厂的人工智能实验室负责人在接受采访时谈到人工智能科学家的流向时曾指出，现在越来越多的企业选择与国内几所大学的学者合作，这样企业和科学家都能有一个相对安全和适当的距离并保证，也能履行各自的职责。

Recently, when talking about the flow of AI scientists in an interview, the person in charge of an AI Laboratory of an Internet factory once pointed out that more and more enterprises now choose to cooperate with scholars from several domestic universities, so that both enterprises and scientists can have a relatively safe and appropriate distance and guarantee, and also can perform their respective duties.

从科学家自身的角度来看，加入企业后存在很多问题。其中之一就是项目已经"提速"。在这一点上，我们只能从国外科学家的情况来类比。

From the perspective of scientists themselves, there are many problems after joining the enterprise. One of them is that the project has been "accelerated". On this point, we can only draw an analogy from the situation of foreign scientists.

有国外媒体曾指出，马斯克创立的明星脑机接口公司neuralink曾经拥有一大批创始科学家。但一些离职的前雇员说，科学的缓慢发展跟不上马斯克苛刻的时间表。

A foreign media once pointed out that neuralink, a star brain computer interface company founded by musk, once had a large number of founding scientists. But some former employees who have left say that the slow development of science can't keep up with Musk's harsh schedule.

科学家有几周的时间来完成某些项目，而研究需要更长的时间来完成，在公司内部形成了一个"压力锅"般的环境，许多员工感到"完全不知所措"。

Scientists are given a few weeks to complete certain projects, while research takes longer to complete, creating a "pressure cooker" like environment inside the company, and many employees feel "completely overwhelmed".

根据最新的报告，大多数公司没有给予足够的时间容忍度，研发也有成本压力。"如果短期内看不到经济效益，它的发展就不是那么快。另一方面，没有足够的人。现有的招聘目标有限。高级管理层还要平衡各个部门，内部其他部门的人员和资源不能完全控制和调度。

According to the late report, most companies do not give enough time tolerance, and R & D has cost pressure. "If we can't see the economic benefits in the short term, its development is not so fast.". On the other hand, there are not enough people. The recruitment targets available are limited. The senior management also has to balance various departments, and the personnel and resources of other internal departments can not be fully controlled and dispatched.

当胡嗅与一些AI学术专家交流时，发现他们在企业工作时有一个共同的不适应，那就是承担了越来越多的管理工作。对他们来说，虽然管理工作不难胜任，但却违背了他们的初衷。

When Hu olfactory communicated with some AI academic experts, they found that they had a common maladjustment when they worked in the enterprise, that is, they took on more and more management work. For them, although the management work was not difficult to be competent, it went against their original intention.

其中一位人工智能科学家表示，在企业中，虽然科学家或科研院所不必直接承担收入的压力，但作为相关业务的负责人，科学家还需要处理业绩核算、业务评估等工作，这就要求科学家在保证科学研究的同时，明确每个人的发展和贡献，从而使原本艰苦的工作更有艰辛的动力。

One of the AI scientists said that in an enterprise, although scientists or research institutes do not have to bear the pressure of revenue directly, as the person in charge of relevant business, scientists also need to deal with performance accounting, business evaluation and other work, which requires scientists to be clear about everyone's development and contribution while ensuring scientific research, so as to make the original painstaking work more painstaking power.

此外，学术科学家长期生活在象牙塔中，很少有人能经得起残酷的商业考验。

In addition, academic scientists have lived in ivory towers for a long time, and few of them can withstand the cruel commercial test.

以中国的一只独角兽为例。在公司的创始团队中，科学家杨帆同时也是公司的技术总监。一开始，团队成员基本上是各司其职。但随着公司的发展和整个行业竞争的加剧，技术的落地和实现逐渐成为公司发展的重点。

Take an autonomous unicorn in China as an example. In the founding team of the company, Yang Fan, a scientist, is also the technical director of the company. At first, the team members basically perform their own duties. However, with the development of the company and the aggravation of competition in the whole industry, the landing and Realization of technology has gradually become the priority of the company's development.

于是，矛盾开始出现。

As a result, contradictions began to appear.

作为技术从业者，杨帆坚持研发的初衷，固执地认为技术达不到一定的极限水平，这还不足以谈落地和实现。这与公司快速销售的想法背道而驰。不久，冲突爆发了。由于技术要求，杨帆与老板发生了几次挑战和争吵。

As a technical practitioner, Yang Fan insists on the original intention of R & D, and stubbornly believes that technology does not reach a certain extreme level, which is not enough to talk about landing and realization. This is contrary to the company's idea of rapid sales. Before long, conflicts broke out. Yang Fan had several challenges and quarrels with his boss because of his technical requirements.

随后，公司里的学术人才开始失去关注，杨帆也开始受到内部压制。首先，有人质疑他管理能力差，然后减少了他的经理人数。从最初管理200人到300人，他被完全提升，相当于坐在板凳上。

Then, the academic talents in the company began to lose their attention, and Yang Fan also began to be suppressed internally. First, he was questioned about his poor management ability, and then he was reduced in the number of managers. From the initial management of 200 to 300 people, he was completely elevated, which was equivalent to sitting on the bench.

当然，故事的最后，杨帆选择了离开，进入了一家新力汽车制造公司。一个小细节是，杨帆离开公司后，得到了同行或同事更多的"祝贺"。很少有人同意"离开是更好的选择"。

At the end of the story, of course, Yang Fan chose to leave and entered a new force car making company. One small detail is that when Yang Fan left the company, he got more "Congratulations" from his peers or colleagues. It's very rare that everyone agreed that "leaving is a better choice".

一般来说，企业与人工智能科学家之间的"分手"更多的是相互选择的结果。

Generally speaking, the "breakup" between enterprises and AI scientists is more the result of mutual choice.

结局：谁能带着尊严离开

The end: who can leave with dignity

因此，人工智能科学家开始离开数年。

As a result, AI scientists began to leave for several years.

例如，贾嘉雅创办的思慕科技，是一家利用AI视觉技术进入智能制造业的公司；百度研究院原院长林元庆创办的艾蜂，是一家致力于垂直产业转型升级的AI集成解决方案公司；飞步科技，由滴滴研究院前院长何小飞创办，是一家专注于无人驾驶卡车领域的公司。

For example, simou technology, founded by Jia Jiaya, is a company that uses AI vision technology to enter the intelligent manufacturing industry; aibee, founded by Lin Yuanqing, former president of Baidu Research Institute, is an AI integrated solution company aiming at the transformation and upgrading of the vertical industry; feibu technology, founded by he Xiaofei, former president of didi research institute, is a company that focuses on the scene of driverless trucks.

但这些离开大企业的企业家真的比以前好吗？答案不确定。

But are these entrepreneurs who leave big business really better than before? The answer is not sure.

不会永远在骚乱中。

Not get forever in the commotion.

一位重返学术界的人工智能学者说，几乎每个研究人员都有一个创业梦想，因为创业的最终目标是影响技术。

An AI scholar who has returned to academia said that almost every researcher has an entrepreneurial dream, because the ultimate goal of entrepreneurship is to influence technology.

不可否认，人工智能科学家离开企业是企业和个人共同选择的结果。当人工智能科学家能够影响的业务范围越来越窄，当企业的竞争回归到商业价值的后半部分时，或许这是双方最经济的选择。

It is undeniable that the departure of AI scientists from enterprises is the result of the choice of both enterprises and individuals. When the scope of business that AI scientists can influence becomes narrower and narrower, and when the competition of enterprises returns to the second half of commercial value, perhaps this is the most economical choice for both parties.

结束语：科学家和企业之间的差距是什么

End: what is the gap between scientists and enterprises

如何弥合科学家与企业之间的鸿沟？

How to bridge the gap between scientists and enterprises?

首先，可以肯定的是，任何科学家的初衷都是把技术落到实处，扩大其影响力。

First of all, it is certain that any scientist has the original intention of putting technology on the ground and expanding its influence.

但是，这种需求只能满足少数企业在一定时期内的需求。例如，智能音箱兴起时，来自百度、阿里巴巴和腾讯的人工智能科学家和自然语言处理专家。例如，当人工智能愿景兴起时，所有公司都在争夺技术巨头。比如现在各大企业都在争夺芯片行业的专家学者。

However, this demand can only meet the needs of a few enterprises in a certain period. For example, AI scientists and NLP (natural language processing) experts from Baidu, Alibaba and Tencent when smart speakers were on the rise. For example, when AI vision was on the rise, all companies were competing for the technology magnates. For example, now all major enterprises are competing for the chip industry experts and scholars.

因此，对于科学家来说，选择合适的时间"出入"是非常重要的。

Therefore, it is very important for scientists to choose the right time to "get in and out".

此外，虽然这些学者可以借着科学家的光环获得更高的薪水，但在一定程度上也意味着更多的放弃。

In addition, although these scholars can get higher salaries with the aura of scientists, it also means more giving up to some extent.

从科学家的角度看，首先要面对的是适应企业，摒弃学术春雪的观念。因此，应作出更多努力，以满足或符合公司的发展需要。

From the point of view of scientists, the first thing to face is to adapt to the enterprise and abandon the concept of academic spring and snow. Accordingly, more efforts should be made to meet or conform to the development needs of the company.

一位同时拥有企业事业部、企业研究院和大学研究院经验的学者表示，这三类院系对于学术人才的优势和劣势非常明显。

A scholar with the experience of enterprise business department, enterprise research institute and University Research Institute at the same time said that the advantages and disadvantages of these three kinds of departments for academic personnel are very obvious.

单从学术研究的角度看，其实技术的影响是进步的。例如，在企业的业务部门，技术只能影响一个或一系列产品，而企业研究所的技术则影响整个企业的技术走向。在高校科研院所中，技术集成和产业技术问题得到越来越多的解决，影响也逐渐扩大。

From the perspective of academic research alone, in fact, the influence of technology is progressive. For example, in the business department of an enterprise, technology can only affect one or a series of products, while the technology of Enterprise Research Institute affects the technology trend of the whole company. In the research institutes of colleges and Universities, the problems of technology integration and industrial technology are solved more and more, and the influence is gradually widening.

然而，上述观点并未得到普遍接受。有人认为高校解决的不是工业技术问题。相反，企业的科研院所似乎是产学研结合的产物。

However, the above views have not been generally accepted. Some people think that what colleges and universities solve is not the problem of industrial technology. On the contrary, the research institutes of enterprises seem to be the combination of industry, University and research.

众所周知，在我国，长期以来，科研普遍是高校的"任务"。这也决定了高校的产业研究更加"应试化"，大多是为了学术竞争和论文发表。高校学者更多的是习惯于在象牙塔里做研究，他们的企业落地能力很弱。

As we all know, in China, for a long time, research is generally the "task" of colleges and universities. This also determines that the industrial research of colleges and universities is more "exam oriented", most of the purpose is for academic competition and paper publishing. Scholars in Colleges and universities are more used to doing research in the ivory tower, and their ability in business landing is very weak.

而这种趋势也延续到了早期的人工智能创业者身上，因为在一些有大学背景的创业者看来，在顶级期刊上发表论文或赢得学术竞赛是快速成长的必由之路。他们更注重证明自己而不是登陆。

And this trend also continued to the early AI entrepreneurs, because in the view of some entrepreneurs with college background, publishing papers in top journals or winning academic competitions is the only way to grow rapidly. They focus more on proving themselves than landing.

事实上，这暴露了我国产学研的一个共同问题，甚至成为人工智能初创企业在一段时间内举世无双如此哀伤的一个重要原因——单凭技术，产品很难被市场认可。

In fact, this exposed a common problem of industry university research in China, and even became a major reason why AI start-ups were so sad all over the world for a period of time - technology alone made it difficult for products to be recognized by the market.

中国生产、教学和科研的现状也使得越来越多的大公司开始寻求与国外大学的合作。例如，华为每年在海外有大量的生产、教学和研究费用，以确保其在技术前沿的优势。

The current situation of production, teaching and research in China also makes more and more large companies start to seek cooperation in foreign universities. For example, Huawei has a large amount of production, teaching and research expenses overseas every year to ensure its advantages in the technological frontier.

从公司的角度来看，初创企业的科学家可能承担更多的责任，比如资源、收入和技术。在大型企业中，受企业内部管理的制约，科学家的行为容易受到约束，最终导致行为的变形。因此，企业在招聘这些人才时，要因材施教，合理分工，发挥优势。

From a corporate perspective, scientists in start-ups may take on more responsibilities, such as resources, revenue and technology. In large enterprises, subject to the internal management of the enterprise, the actions of scientists are easy to be restrained, which eventually leads to the deformation of actions. Therefore, when recruiting these talents, enterprises should teach students in accordance with their aptitude, reasonably divide responsibilities and give full play to their strengths.

但归根结底，这实际上是企业和科学家希望从人工智能浪潮中得到更多，也希望彼此得到更多。毕竟，没有一个企业是一个可以免费支付的慈善组织。相反，资本的本质是少花钱多办事。

But in the final analysis, it's actually that enterprises and scientists want more from the wave of AI and more from each other. After all, no enterprise is a charity organization that can pay at no cost. On the contrary, the essence of capital is to spend less and do more.

这是科学家将继续做出职业选择和价值衡量的重要原因之一。

This is one of the important reasons why scientists will continue to make career choices and value measurement.

可见，科学家和企业之间的差距并不难跨越，但企业和个人总是在做一个关于"成本效益"的选择。

So it can be seen that the gap between scientists and enterprises is not difficult to cross, but enterprises and individuals are always doing a choice about "cost performance".

我们可以预见，在未来，人工智能科学家将重返学术界，否则创业的趋势将继续。

We can foresee that in the future, AI scientists will return to academia, or the trend of entrepreneurship will continue.

Link：https://new.qq.com/omn/20210301/20210301A01H0J00.html

update time：2021-03-01 10:41:45