Have you learned about the opportunities and challenges brought to the country by the era of artific

 

1. Artificial Intelligence Image Robot

With the accelerated evolution of technological progress and industrial transformation, artificial intelligence (AI) has become a competitive technological innovation highland for countries. Looking at the world, AI has been highly valued at various levels such as government, academic institutions, and enterprises. Its development in academic research, technological innovation, talent education, and other aspects has shown a new trend.

2. Artificial intelligence image generation

01 Global Development Situation 1.1 Academic Research More and more countries are paying attention to AI research, and AI has become a hot topic in the academic field. According to Microsoft Academic Graph (MAG) statistics, from 2000 to 2020, the number of AI papers published through various channels increased from less than 48000 to 230000, an increase of about four times (see Figure 1), and the proportion of all papers increased from less than 2% to 3%.

3. High definition artificial intelligence image materials

The "Artificial Intelligence Index Report 2021" shows that East Asia, Europe, and North America are the main regions for paper production in each major country and region, with universities being the largest contributors to paper production.

4. Artificial intelligence image creativity

But the second place is different: in the United States, the second place is enterprises, accounting for about 19.2% of all papers, while in China and the European Union, the second place is public research institutions, accounting for 15.6% and 17.2%, respectively

5. Artificial intelligence image materials

Scientific research journals are still the main carrier for paper publication. In 2020, the number of AI papers published in journals was nearly 80000, 5.4 times that of 2000. Especially in recent years, the growth rate has been significant, with an increase of 19.6% in 2019 compared to 2018, and a further increase of 34.5% in 2020 compared to 2019.

6. Artificial intelligence image high-definition

East Asia, Europe, and North America are the top three regions with the highest number of journal publications in the AI field. Since 2017, China has been the country with the highest number of journal publications in the AI field, and in 2020, it surpassed the United States for the first time to become the country with the highest frequency of cited papers in the world's artificial intelligence journals, accounting for 20.7%, while the United States was 19.8%.

7. Artificial Intelligence Picture Cartoon

Academic conferences are another important way of publishing papers. From 2000 to 2019, the number of articles published through AI academic conferences increased fourfold, but the growth was slow from 2010 to 2020, maintaining around 40000 papers. The number of conference papers in 2019 was only 1.09 times that of 2010, and due to the impact of the epidemic, it was reduced by more than 20% compared to 2019.

One of the methods for artificial intelligence images is to use object detection

1.2 Technological Innovation 1. High intensity AI research and development has produced a large amount of scientific research results. Apart from a slight decrease in the impact of the epidemic in 2020, the number of AI patents has experienced a rapid increase since 2000 (see Figure 2). According to incomplete statistics from Stanford University on more than 40 countries worldwide, the number of AI registered patents worldwide in 2019 has reached 101876, which is 4.7 times higher than 21806 in 2000, The proportion of all patents has also increased from 2.0% in 2000 to 2.9%.

One of the methods for artificial intelligence film reading is to use object detection

From a regional perspective, the main production area for AI patents is North America, with approximately 55% of patents originating from North America. In addition, Europe and East Asia each account for about 20%

The process of artificial intelligence film reading is reflected in

Secondly, AI technology has further improved. AI systems can now synthesize text, audio, and images at a sufficiently high level, making it difficult for humans to distinguish authenticity. For example, image synthesis technology can "deeply forge" by overlaying faces onto other people's faces in photos or movies. This has also prompted researchers to explore deep forge detection technology, allowing computers to distinguish different image outputs well.

Thanks to the progress of machine learning and natural language processing technology, machines can provide more accurate natural language answers in visual question and answer. Since its first release in 2015, the accuracy rate of this algorithm has increased by nearly 40%, reaching 76.4% at the highest, which is close to 80.8% of the accuracy rate of human beings. The third baseline is to gradually move towards industrialization.

On the one hand, the maturity of technology has significantly reduced the training time and cost of relevant AI models. For example, training a modern image recognition system. According to testing conducted by the Stanford DAWNBench team, a project that required a cost of $1100 in 2017 now only costs $7.43, with a cost of only 1/150 of the original cost.

On the other hand, AI has been applied in an increasing number of fields, such as transportation, finance, agriculture, and military, gradually becoming more "intelligent". Especially during the pandemic, with the adoption of machine learning, the pattern of the healthcare and biomedical industries has undergone substantial changes, greatly simplifying the original compound structure design technology.

The machine learning technology of AI startup PostEra can complete the chemical synthesis route design that required 3 to 4 weeks in 48 hours, accelerating the discovery of COVID-19 related drugs; The deep learning model AlphaFold, originally developed by Google's artificial intelligence department DeepMind, has solved decades of protein folding biology challenges.

The fourth is a significant increase in investment. In 2020, the global total social investment in the AI field (including private investment, public offerings, mergers and acquisitions, and minority equity) increased by 40% compared to 2019, reaching 67.9 billion US dollars. From a national perspective, the United States remains the region with the most concentrated capital. In 2020, social capital investment in AI exceeded 23.6 billion US dollars, followed by China (9.9 billion US dollars) and the United Kingdom (1.9 billion US dollars).

From a field perspective, the biopharmaceutical sector received the most investment in 2020, exceeding 13.8 billion US dollars, which is 4.5 times that of 2019; Secondly, in the field of smart transportation, it reached 4.5 billion US dollars; Another $4.1 billion in the education sector, however, the number of AI startups receiving investment has significantly decreased for three consecutive years. In 2017, over 4000 AI startups achieved financing, and by 2020, there were only less than 1000.

1.3 Talent and Education In 2020, the number of talent recruitment in the global AI field continued to grow. In the past five years, Brazil, India, Canada, Singapore, and South Africa were the top five countries with the fastest growth in the proportion of AI positions in new positions. For example, in Brazil, the proportion of AI positions increased more than three times in the past five years, while in China, the growth rate was relatively slow, only 1.3 times; The United States has even experienced its first decline in years, with AI jobs announced decreasing from 325000 in 2019 to 301000 in 2020.

In terms of education, top universities around the world have deployed to increase investment in AI education. Between 2017 and 2020, the number of AI related courses for undergraduate students increased by 102.9%, while for graduate students, it increased by 41.7%. At the same time, more and more AI doctoral students are choosing to leave the school and enter the industry. From 2010 to 2019, the proportion of fresh doctoral graduates entering enterprises to work in AI related fields increased from 44.4% in 2010 to 65.7% in 2019, The proportion of people continuing to engage in academic research has decreased from 42.1% to 23.7%.

Taking the United States as an example, among those who obtained doctoral degrees in computer science, the proportion of AI related doctoral students was 14.2% in 2010, and increased to about 23% in 2019. At the same time, the popularity of other computer science fields has decreased, such as networks, software engineering, and programming languages. In 2019, the proportion of international students among newly enrolled AI doctoral students continued to rise, reaching 64.3%, an increase of 4.3% compared to 2018.

Among foreign graduates, up to 81.8% choose to stay in the United States. The rapid development of AI technology in the future is due to its wide range of application scenarios. AI is no longer just a simple software or hardware, but has become a collection of elements including algorithms, data, hardware, applications, talents, etc. With the rapid improvement of computer system problem-solving and task execution capabilities, Machine intelligence is increasingly replacing human intelligence in more and more fields, and AI has become the most powerful tool available to humans, which can expand knowledge, promote social prosperity and prosperity.

The following are key aspects of future AI applications with rich human experience: firstly, prediction. With the advancement of models and algorithms, AI can more accurately predict, summarize and learn from past data, and predict future events. This can be applied in almost all fields, such as weather forecasting, predicting the trajectory of other vehicles in autonomous driving, and making decisions on planting, irrigation, and fertilization in precision agriculture.

The second is design and optimization, which involves optimizing and coordinating a series of complex tasks to achieve the goals of saving time and money, improving safety, etc. according to needs, such as planning and design of smart cities, transportation route planning, etc. The third is modeling and simulation. By constructing virtual models and conducting simulation operations in biological, physical, economic, and social research, many practical tests and experiments can be saved, Provide more solutions while improving experimental efficiency.

In the COVID-19 pandemic, researchers widely used AI technology in drug molecular design, protein conformation and other research. Fourth, natural language processing. The timely and accurate interaction between human natural language and computer can not only make the operation of intelligent electronic device easier, but also can be widely used in text classification, machine translation, public opinion monitoring and other fields.

The fifth is visual image processing, which uses cameras and computers to replace the human eye in machine vision for target recognition, tracking, and measurement, helping to make correct "decisions". AI can be applied in more fields that require perception from images or multidimensional data, such as production line product detection, autonomous driving, medical imaging analysis, etc.

The development characteristics and challenges of 03 have led to the practical application of AI in fields such as scientific research, healthcare, education, and smart cities. AI is ubiquitous in daily life and is reshaping our lives in various ways. (1) The pace of AI innovation is accelerating the updating and upgrading of several hot technologies in the AI field, such as machine learning, computer vision, and robotics.

For example, in the field of computer vision, the accuracy of Google's "Google Brain" model for image recognition was generally between 60% and 70% in 2013, and could reach more than 90% in 2020. The speed of model training has increased by 8 times. For example, in the field of intelligent translation, there were only 8 independent machine translation cloud platforms with training models in the business field in 2017, increasing to 28 in 2020, The error rate for word recognition in the LibriSpeech database has decreased from 5% in 2017 to only 1%.

(2) The dissemination of AI research and development tools has become more widespread. The cutting-edge deep learning technology requires a large amount of data, computing power, and professional knowledge, which is very expensive. With the accelerated progress of technology, AI applications and development tools have gradually become public and available to more people. Many platforms are open source and free, and training costs in many fields have also been significantly reduced.

The popularization of cloud computing and data sharing will make AI innovation no longer exclusive to a few people and can be implemented globally. (3) AI is changing human-machine relationships in modern society, and people's demand for machines and automation is constantly deepening. For example, smartphones nowadays have various AI supporting functions, including voice assistants, photo tagging, facial recognition security, search applications, etc Recommendation and advertising engines, etc.

As the functions of mobile phones become increasingly powerful, people's dependence on them is also upgrading. In addition, smart appliances, smart homes, and others are changing people's living habits by providing more convenient, accurate, and efficient services. AI is transforming past fantasies into interdisciplinary reality, allowing machines to play a more important role. Its development and application are showing a more positive trend, but also facing new challenges.

With the development of AI from elite science to mainstream tools, it may encounter bottlenecks that are difficult to break through in some aspects in the future. Firstly, the data required for training models will become more precious, and governments, enterprises, and research institutions around the world will pay more attention to and strengthen the protection of data and privacy, or limit the rapid development of AI; Secondly, with the popularization of applications, AI will face practical ethical challenges in employment, biology, social equity, and even the relationship between machines and humans; The third is the disruptive innovation brought by AI in some fields, such as autonomous driving. Whether relevant regulatory measures can be followed up in a timely manner will have a profound impact on its application.

04 The global competition pattern is precisely because of its broad application prospects. AI is expected to become an important factor in shaping the global competition pattern, or will bring significant economic and strategic advantages to early adopters. Many countries are strengthening AI R&D and application. AI competition has gradually changed from competition between research institutions and enterprises to competition between countries.

4.1 Countries actively participate in AI competition. Since 2017, more than 30 countries and regions around the world have issued national strategies to prioritize the development of AI and have joined the global competition for AI (see Table 1), including developed countries and regions such as the United States, the European Union, France, Germany, the United Kingdom, Singapore, as well as developing countries such as India, Mexico, Indonesia, Brazil, Malaysia, Ukraine, etc.

All countries are attempting to support the development of their AI industry through AI strategies, occupying a favorable position in international competition

Canada, Germany, India, the United Kingdom, and other countries have clearly stated in their strategies that they need to provide direct financial support for the development of AI, and have listed budget amounts. Countries such as Japan, the United Kingdom, and Saudi Arabia have also established specialized AI management agencies with strong policy support, and their AI technology, talent, and industry are all developing rapidly.

Singapore has become the country with the highest proportion of AI positions. According to 2020 data, the proportion of AI related positions in all job positions reached 2.4%. According to statistics, among the top 50 skill positions in India, the number of AI technology related positions is the highest, even surpassing AI giants such as the United States, China, and Germany. 4.2 China and the United States are leading the way

Among many countries, the United States is still recognized as the first AI power, with strong enterprises and research institutions, and the government also invests a lot of financial resources to support research and development every year. In fiscal year 2020, the AI allocation of non defense and intelligence departments will reach 973.5 million dollars, and in fiscal year 2021, this number will further increase to 1.5 billion dollars, an increase of nearly 55% year-on-year.

However, the leading position of the United States is gradually catching up with China, and some fields have even been surpassed. For example, China has maintained the top number of global AI paper publications since 2017 and gradually expanded its leading advantage. In 2020, it surpassed the United States for the first time to become the country with the highest citation frequency in AI journals in the world.

In 2019, the Data Innovation Center of the Information Technology and Innovation Foundation (ITIF) in the United States released a research report titled "Who Will Win in the Artificial Intelligence Competition: China, the European Union, or the United States?" to compare and calculate the current development status of AI in China, the United States, and Europe - the United States leads by 44.2 points, China ranks second by 32.3 points, and the European Union ranks third by 23.5 points.

The leading position of the United States is undoubtedly evident, while China is catching up with former Google Chairman Eric Schmidt. He believes that China has a leading position in facial recognition, health data, e-commerce, and many other areas. The United States is "only one to two years ahead of China" in the development of AI, and China's rapid rise has caused concern from all sectors of the United States.

A large number of American media and think tanks advocate that China is about to surpass the United States in the field of AI. On March 4, 2021, the National Artificial Intelligence Security Council (NSCAI) of the United States voted and submitted a 756 page final recommendation report to the US Congress, detailing the important role of AI in social development and national security, and proposing measures and suggestions on how to maintain AI technology advantages and ensure national security, In particular, China will be identified as the largest competitor of the United States in the AI field in the future, and the federal government of the United States will be targeted to take measures.

The report has become a bipartisan consensus between the Democratic and Republican parties in the United States and has received attention from various sectors, including the federal government. The Biden administration has also expressed support for many specific policy recommendations mentioned in the report in areas such as intellectual property, talent, accelerated innovation, microelectronics, and international cooperation. It is foreseeable that the strategic game in the AI field between China and the United States will further evolve in the future, and competition in attracting talent and breaking through key technologies will become more intense.

4.3 International Cooperation Actively With more and more countries participating in global AI competition, there is an urgent need to develop global practices有效的国际规则、标准，同时国家间也需要通过互利合作达到取长补短、共同进步的目的为满足一些国家共同发展的目的，各种AI倡议应运而生，很多非常重要的国际倡议、论坛、组织也纷纷纳入AI议题，涉及AI的原则标准、数据共享、研发合作等方面内容（见表2）。

日本、韩国、英国、美国和欧盟成员国等国家积极参与了政府间AI方面的合作也有一些国家通过积极缔结双边协议来推进AI国际合作

值得注意的是，不同于特朗普的一意孤行，拜登政府更加注重通过联合盟友维护其世界霸主的地位专家预测美国或将在科技领域牵头发起成立“科技民主联盟”，借西方民主价值观拉拢盟友共同促进AI和新兴技术的发展和使用，并进一步对中国实施科技封锁与遏制，中美科技竞争或将升级迈入新阶段。

05有关建议虽然我国AI技术研发在全球的影响力显著提高，又拥有全球最丰富的应用落地场景，但AI伦理、人才培养等问题仍需重视习近平总书记在党的十九大报告中明确提出，要“推动互联网、大数据、人工智能和实体经济深度融合”，将人工智能视为供给侧结构性改革和推动实体经济发展的重要一环。

相关部门贯彻落实十九大精神，出台了一系列战略举措，有力地推动我国在AI技术和产业化方面的进步结合近年来全球AI发展动态，为贯彻落实新发展理念，进一步加快我国AI技术创新，提出以下建议：（1）理顺促进AI发展的体制机制。

为更好地发挥政府在AI创新方面的重要作用，一些国家成立了指导AI发展的专门机构，一方面解决部门间各自为战，碎片、重复支持AI研发的情况，另一方面也可更高效地加强对民间AI创新的引导建议进一步理顺我国促进AI发展的体制机制，成立专门机构或小组负责统筹部门间的支持政策，避免重复支持、无效支持，实现政府资源效益最大化和跨部门政策的良好衔接，并进一步促进AI成果转化，使我国从AI论文强国变成AI专利强国。

（2）强化AI数据资源共享近一两年来，鼓励相关数据和模型的资源共享已经成为美国、英国、日本、韩国等国家在AI管理上的重要共识，也是国际合作的一项重要议题这是由AI技术的特征决定的，为了朝更智能的方向前进，最基础、最关键的大数据支撑是必不可少的。

美国谷歌、Meta（原脸书）、亚马逊等科技企业凭借其商业模式和垄断地位掌握了大量用户数据，对其进一步研究机器学习、深度学习至关重要我国在加强数据保护的基础上，要充分利用数据资源，鼓励开放共享，推进行业整体发展前进。

（3）加强人才培养人才是第一资源，在未来AI战略竞争中，谁能够在世界范围内吸引、培养和留住人才，谁就会取得优势据统计，我国AI人才目前缺口超过500万，国内的供求比例为1∶10，供需比例严重失衡，同时80%以上的外籍美国AI博士应届毕业生选择留在美国。

继续加强人才培养、补齐人才短板，是我国的当务之急特别要探索产学合作、国内国际合作、跨界跨领域合作的育才机制，营造有利于AI人才成长与培养的沃土，推动构建和完善既有利于发展人才独创能力，又能有效调动潜力的平台条件，为未来做好准备。

（4）营造有利的国际环境AI未来发展之路还有很长，以美国为首的发达国家一方面以举国之力发展本国AI技术，另一方面也在充分发挥其影响力积极参与AI国际规则的制定，甚至意图联合盟友对我国进行战略打压凭借现有产业基础和技术实力，我国有能力且应当在未来AI领域竞争中发挥比较优势，补齐发展短板，下好先手棋，努力成为行业规则的重要制定者，引领技术朝有利于我国的方向发展，逐步在新的国际体系中占据重要位置，实现社会主义现代化强国的建设目标。

（5）重视AI发展的社会伦理和法规建设AI在多个领域的加速应用，导致相应的伦理问题逐渐凸显，尤其是隐私、就业、社会公平等领域，也是许多国际组织、协会、联盟等积极讨论的议题，甚至可能成为美欧等国打压我国AI产业的关键。

同时，对AI的监管必须与时俱进，在交通、金融、医药等应用较好的领域要及时完善相关行业法律法规，明确发展AI中的伦理规则，积极引导，让新业态有规可循、有法可依，促进应用落地（6）密切跟踪全球发展趋势当前AI创新进入高速发展阶段，相关技术加速进步，产品日新月异，产业爆发式增长。

美国、德国、韩国、以色列等科技强国都结合自身特点选择重点领域布局突破要持续跟踪把握领域前沿发展趋势，掌握发达国家在研发方向、支持举措、监管规则等方面的最新动态，结合我国国情学习借鉴免责声明：本文转自科情智库，原作者程晓光。

文章内容系原作者个人观点，本公众号转载仅为分享、传达不同观点，如有任何异议，欢迎联系我们！推荐阅读2021年世界前沿科技发展态势及2022年趋势展望——综述篇2021年世界前沿科技发展态势总结及2022

年趋势展望——信息篇2021年世界前沿科技发展态势总结及2022年趋势展望——生物篇2021年世界前沿科技发展态势总结及2022年趋势展望——能源篇2021年世界前沿科技发展态势总结及2022年趋势展望

——海洋篇2021年世界前沿科技发展态势总结及2022年趋势展望——航空篇2021年世界前沿科技发展态势总结及2022年趋势展望——航天篇2021年世界前沿科技发展态势总结及2022年趋势展望——新材料篇

2021年世界前沿科技发展态势总结及2022年趋势展望——先进制造篇转自丨科情智库作者丨程晓光选自丨译者丨编辑丨

研究所简介国际技术经济研究所（IITE）成立于1985年11月，是隶属于国务院发展研究中心的非营利性研究机构，主要职能是研究我国经济、科技社会发展中的重大政策性、战略性、前瞻性问题，跟踪和分析世界科技、经济发展态势，为中央和有关部委提供决策咨询服务。

“全球技术地图”为国际技术经济研究所官方微信账号，致力于向公众传递前沿技术资讯和科技创新洞见地址：北京市海淀区小南庄20号楼A座电话：010-82635522微信：iite_er