The road to the future of artificial intelligence China McKinsey report, the report reviews the development of artificial intelligence technology process, and analyzes the technology for Chinese social development may play a role, prediction of related industries and economic policy framework.
1 inflection point approaching
The degree of commercialization in the field of artificial intelligence
AlphaGo successfully defeated the human world chess champion Li Shishi, between the concept of artificial intelligence science fiction seems to be a night from fantasy into reality with no reality whatever.
Traditional software programs are written by humans, including specific instructions. The working mode of artificial intelligence is completely different. They are based on a common learning strategy, you can read a large amount of big data, and found that the law, contact and insight. Therefore, artificial intelligence can automatically adjust according to the new data, without the need to reset the program.
事实上， 20 世纪五六十年代，人工智能的理论就已经初步形成，但由于技术未能实现突破性进展，人工智能无法达成预期效果，因此陷入了一段沉寂期。
In fact, in twentieth Century 50s and 60s, the theory of artificial intelligence has been formed, but due to technical failure to achieve a breakthrough, artificial intelligence can not reach the expected effect, in a period of silence.
往后数十年间虽然不乏成功案例（如 IBM 的超级计算机“深蓝”击败国际象棋冠军加里·卡斯帕罗夫），但因为人工智能在现实世界的成功案例太过孤立，所以不足以支撑大规模商业化。
For several decades although there are successful cases (such as the super computer IBM blue beat chess champion Gary Kasparov), but because the success of artificial intelligence in the real world is too isolated, so it is not enough to support large-scale commercial.
现在，我们正处于人工智能复兴浪潮，数据收集及整理 、 算法以及高性能计算等技术的突飞猛进促成了革命性进步。利用机器学习，人工智能系统获得了归纳推理和决策能力，并实现了某些商业化落地，可以担当客服、管理物流、监控工厂机械、优化能源使用以及分析医学资料；而深度学习更将这一能力推向了更高的层次。
Now, we are in the revival of artificial intelligence, data collection and collation, algorithms and high-performance computing technology, such as the rapid progress of the revolutionary progress. The use of machine learning, artificial intelligence system to obtain the inductive reasoning and decision-making ability, and to achieve a certain commercial floor, can serve as the customer service management, energy use and analysis of medical data, factory machinery, logistics monitoring and optimization; deep learning will be more this ability to a higher level.
Change not only in the theoretical front. As a pioneer in the future of super intelligent systems – analytical tools for various types of machine learning technologies have emerged. Financial, medical, manufacturing and other industries, the rapid development of the application of artificial intelligence in the field of global venture capital soared from $589 million in 2012 to $about 5000000000 in 2016. McKinsey predicts that by 2025 the total market value of artificial intelligence applications will reach $127 billion.
In the future, artificial intelligence can become a powerful tool to deal with some of the core challenges of society. In the medical field, artificial intelligence will greatly enhance our ability to analyze the human genome and develop personalized treatment programs for patients, and even greatly accelerate the process of curing cancer, Alzheimer’s disease and other diseases. In the field of environmental protection, artificial intelligence can analyze climate characteristics and reduce energy consumption in a large scale. Artificial intelligence can even play a role in the area outside of the earth, or help the future human exploration of Mars and the outer space.
More than 2000 contents of a recent report by the McKinsey Global Institute covers 800 of the world’s various occupation analysis found that about 50% of the world’s work can be achieved by improving the existing technology of automation. Of course, the technical feasibility is only one factor affecting the speed and degree of automation, there are other factors to consider, including the research and application of cost, the labor market supply and demand, economic benefits, social acceptance, and government supervision department.
The above factors, the Automation Research Report by the McKinsey Global Institute pointed out that at present all the work content, half automation in about 2055, but this process there are many variables, the error in about 20 years.
2 Chinese has become one of the world’s AI Center for the development of the academic field, only in 2015, two related papers published in academic journals totaling nearly 10 thousand copies, while the sum of the academic research in Britain, India, Germany and Japan published is only equivalent to half of the. Among them, the development of China’s artificial intelligence is driven by the leading enterprises of science and technology, such as automated personal assistant, automatic driving cars, etc.. This is due to a large number of search data and a wealth of product lines, some Internet companies in the forefront of Natural Language Processing, image and voice recognition technology. But it is worth noting that, although the number of papers in the field of artificial intelligence in China than the United States, but the influence of Chinese scholars is still less than the United States or British counterparts.
The number and influence of Chinese artificial intelligence
Artificial intelligence ecosystem, the United States is also more perfect and active, far more than the number of start-up companies in china. By the research institutions, universities and private enterprises constitute a huge ecosystem, innovation and diversity. Silicon Valley in the field of science and technology accumulated strong strength to form a strong and difficult to replicate the advantages.
The United States AI ecological more perfect and active
In addition to technical reserves and entrepreneurial ecology, China in the development of artificial intelligence, there are two issues to consider:
Stopped in the mountains of China AI Road
Data environment to be released
Chinese government data openness of the world’s ninety-third
First of all, while China’s technology giants can get huge amounts of data through its proprietary platform, but in the creation of a unified standard, cross platform sharing of data friendly ecosystem, China still lags behind the United states. Second, countries around the world have come to realize that open government databases help to promote innovation in the private sector, but the openness of the Chinese government data is still very limited. Finally, the restrictions on cross-border data flow also make China in a disadvantageous position in global cooperation.
Talent market tension
应用层面而言，中国的算法发展程度与其他国家并无太大差距。然而，中国的研究人员在基础算法研发领域仍远远落后于英美同行。一个主要原因就是人才短缺。美国半数以上的数据科学家拥有10年以上的工作经验，而在中国，超过 40%的数据科学家工作经验尚不足5年 。中国在人才方面的持续努力将至关重要。
Application level, the development of China’s algorithm is not much difference with other countries. However, Chinese researchers in the field of basic algorithm development is still far behind the British and American counterparts. One of the main reasons is the shortage of talent. More than half of the data scientists in the United States have more than 10 years of work experience, while in China, more than 40% of the data scientists work experience is less than 5 years. China’s continued efforts in talent will be crucial.
目前，中国只有不到 30 所大学的研究实验室专注于人工智能，输出人才的数量远远无法满足人工智能企业的用人需求。此外，中国的人工智能科学家大多集中于计算机视觉和语音识别等领域，造成其他领域的人才相对匮乏。如果中国大学对学生提出更高的数学和统计学要求，并且集中资源发展该领域全球前沿研究，人工智能的发展必将受益匪浅。另一个值得思考的方向是改进现有的科研经费分配模式来推进创新。
Hardware short board
High computing speed is the most important technology in the development of cutting-edge artificial intelligence technology; special processors, such as GPU can handle a large number of complex computing, the development of artificial intelligence is particularly important. And its energy consumption for a long time, China’s heavily dependent on imports of microchips, some types of high-end semiconductor is almost entirely dependent on the level of imports determines whether artificial intelligence solutions to achieve large-scale commercialization.
2015 年，美国政府禁止了英特尔、英伟达和 AMD这三家全球最大的芯片供应商向中国机构出售高端超级电脑芯片。这一禁令显示了中国在半导体方面的自主研发能力对于未来人工智能发展十分重要。
In 2015, the U.S. government banned Intel, NVIDIA and AMD three of the world’s largest chip supplier to sell high-end super computer chip to Chinese mechanism. The ban shows that China’s independent research and development capabilities in the semiconductor is very important for the future development of artificial intelligence.
为应对这一局面，中国政府在 2014 年出台了《国家集成电路产业发展推进纲要》以及“中国制造 2025”行动纲领。中国政府还成立了国家集成电路产业投资基金，目前募资已超过 200 亿美元。相关行动已初见成效：2016 年 6 月神威?太湖之光超级计算机问世，成为世界上运算速度最快的超级计算机，使用的是中国自主知识产权的处理器。政府的前期投资可以产生显著的涟漪效应，鼓励私营企业的积极参与。
In response to this situation, the Chinese government in 2014 issued a national outline for the development of integrated circuit industry, as well as the made in China 2025 action plan. The Chinese government has also set up a national integrated circuit industry investment fund, currently raising more than $20 billion. The relevant action has been noticed: June 2016 Taihu light Kamui? The advent of super computer, super computer became the world’s fastest, is the use of Chinese independent intellectual property rights of the processor. The initial investment of the government can produce significant ripple effect and encourage the active participation of private enterprises.
Complex social and economic problems
Labor market demand and impact
在过去数十年，中国因“人口红利”受益良多，劳动力的扩张大大促进了经济增长。但老龄化正使中国逐渐失去这一推动力。中国的劳动年龄人口最早将在2024 年达到峰值，并在之后的 50 年中减少五分之一。
In the past few decades, China has benefited greatly from the demographic dividend, which has greatly promoted economic growth. But ageing is making China lose its momentum. China’s working age population will peak in 2024, and by 1\/5 in the next 50 years.
This demographic change means that, on the basis of the current level of productivity, China will lack enough labor to sustain its economic growth. The only practical way to boost economic growth is to boost productivity growth, such as the introduction of artificial intelligence. According to the McKinsey Global Institute, based on the speed of the application of artificial intelligence, automation brings productivity gains for China to contribute 0.8 to 1.4 percentage points per year of economic growth.
Artificial intelligence has the potential to significantly increase productivity growth, but the price may be further widening the income gap, including the imbalance of urban development. All in all, artificial intelligence will drive the formation of what is known as skill – oriented technological change – that digital skills will be of particular importance, and that demand for skilled workers will be reduced.
中国目前从事可自动化工作的劳动力人口超过其他国家。麦肯锡全球研究院预测中国 51%的工作内容有自动化潜力，这将对相当于 3.94 亿全职人力工时的冲击。但这并不意味着如今的高端工种能够完全免受冲击。比如，医生之类专业人士的部分工作也可能被自动化，而医生的工作内容将会更专注于与人的沟通和互动。
China is currently engaged in the work of automation more than other countries. McKinsey Global Institute predicts that 51% of China’s work content has the potential for automation, which will be equivalent to the impact of the equivalent of 394 million full-time human hours. But that does not mean that today’s high-end jobs can be completely free from shocks. For example, part of the work of a professional such as a doctor may also be automated, and the doctor’s work will be more focused on communication and interaction with people.
Social benefits and worries
Artificial intelligence development prospects, can be used to improve health care, environment, safety and education, improve people’s well-being. At the same time, due to the blurring of physical reality, digital and personal boundaries, complex ethical, legal, and security issues are derived. With the gradual popularization of artificial intelligence, the need for careful management to deal with this change.
Many of the existing cases show artificial intelligence to solve social problems such as environmental change, forecasting potential, energy saving and emission reduction, intelligent medical, digital medical records, more safe and efficient transportation system, intelligent city and automatic driving.
However, in addition to these potentials, the management of independent learning and decision-making ability of the machine is also a heavy responsibility.
First of all: who owns personal data? How should data be shared? Faced with the increasingly serious network security attacks and how to protect the data? Secondly, artificial intelligence may produce unconscious discrimination in the decision-making process. Because of the existence of various forms of racial discrimination, gender discrimination and Prejudice in the real world, the data in the input algorithm may also be accompanied by these features. And when the machine learning algorithm is trained on these biased training data, it inherits the bias. There are also a number of geopolitical implications, such as the digital divide between nations, automated weapons, etc..
In addition to ethical issues, the popularity of artificial intelligence in the community will have a lot of impact on the legal level. Who should be responsible for the artificial intelligence that leads to accidents and even crimes? Who is the intellectual property of artificial intelligence? Once artificial intelligence has super power, and what measures should be used to monitor? What are the legal rights and obligations of R & D personnel?
To establish a sound legal and ethical framework, there are still many issues to be fully explored.
Five prediction of China’s AI strategic focus
Establish a sound data ecosystem
Data is the currency of the future. For example, in medical research, without the support of the world’s vast clinical data, the potential of artificial intelligence can not be fully excavated. Too many shackles will be bound by China’s artificial intelligence business, resulting in the loss of the ability to develop products with global competitiveness.
China can build a more complete data ecosystem by establishing and implementing data standards, opening public data to the private sector, and encouraging cross-border data exchange.
首先， 建立 数据标准是进行广泛数据分享和实现系统间交互操作的重要前提条件，有助于提升物联网及人工智能技术的价值。潜在的庞大数据体量是中国的天然优势，使中国有机会在国际上更好地发挥领头羊的作用。而且，在与中文语言相关的数据规范制定方面，中国也应起到主导作用。
First of all, the establishment of data standards is an important prerequisite for a wide range of data sharing and interoperability between systems, and help to improve the value of things and artificial intelligence technology. Potential huge data volume is China’s natural advantage, so that China has the opportunity to play a better role in the international leader. Moreover, China should play a leading role in the development of Chinese language related data specification.
对于特定行业数据，政府可要求现有的监管机构制定必要规则。比如美国证券交易委员会在 2009 年出台规定，要求所有上市公司使用 XBRL（可扩展商业报告语言）格式发布财报，确保所有公开数据的机器可读性。
For specific industry data, the government may require existing regulatory agencies to develop the necessary rules. For example, the U.S. Securities and Exchange Commission in 2009 issued a requirement for all listed companies to use XBRL (Extensible Business Reporting Language) format to publish earnings, to ensure that all the data of the machine readable.
其次，为了提升数据的多样性，政府应提高公共数据的开放程度 ，并带头建设行业数据库。这些举措同时能够提升公共服务质量、提供政策制定洞见，从而带来额外益处。比如纽约市政府就建立了公开数据门户网站，为市民提供经济发展、医疗、休闲、公共服务等领域的数据。2012 年纽约市还颁布了《开放数据法案》，要求政府部门使用机器可读取的数据并建立 API（应用程序编程接口），方便软件研发人员直接连接政府系统并获取数据。
Secondly, in order to improve the diversity of the data, the government should improve the openness of public data, and take the lead in the construction of the industry database. These initiatives can also improve the quality of public services, provide policy insight, resulting in additional benefits. For example, the New York municipal government on the establishment of an open data portal for the public to provide economic development, health care, leisure, public services and other areas of data. In 2012, New York city has also promulgated the open data act, requires government departments to use the machine readable data and the establishment of API (Application Programming Interface), convenient software R & D personnel directly connected government system and access to data.
最后，中国政府还需考虑 国 际 数据流的价值。麦肯锡全球研究院的调查表明，2014 年，跨境数据流为全球经济创造了 2.8 万亿美元的价值，对经济增长的贡献已经超过实体贸易。此外，研究还指出，由于经济体需要接触全球的思想、研究、科技、人才和最佳实践案例，数据流入和流出都能十分重要 。
Finally, the Chinese government needs to consider the value of international data flow. According to a survey by the McKinsey Global Institute, in 2014, the cross-border data flow created a value of $2 trillion and 800 billion for the global economy. In addition, the study also pointed out that because the economy needs to contact the global thinking, research, science and technology, talent and best practice cases, data inflows and outflows are very important.