
人工智能、数字经济发展浪潮下,算力无处不在,它们之间有着怎样的关系?人工智能发展的终极目标是什么?为抢抓人工智能机遇,我国需要做好哪方面工作?
带着这些疑问,记者对话中国科学院院士、中国数字经济50人论坛委员王怀民,了解人工智能、算力发展浪潮背后的产业逻辑及未来趋势。
王怀民,中国科学院院士,中国计算机学会(CCF)会士,CCF开源发展委员会主任,国家百千万人才工程领军人才,教育部科学技术委员会委员,长期从事分布计算领域研究和人才培养,在分布计算模型、技术与平台方面做出系统性和创造性工作,为我国分布计算核心关键技术自主创新,为国家和军队网络信息系统建设和发展作出突出贡献。作为第一完成人,王怀民院士曾多次获得国家技术发明奖、国家技术进步奖等。
王怀民表示,人工智能是一个人机协作、持续学习、成长演化的新的数字经济生态,它不是让人走开,而是让人和机器更好地合作。
一、算力是数字经济发展的驱动力
Q:在当下数字经济发展的背景下,各界都在提算力,您如何理解算力和数字经济以及人工智能之间的关系?
王怀民:数字经济离不开算力,也正因为计算技术的发展,才有了今天数字经济的可能性,因此算力是形成数字经济认知的技术基础。 CXO UNION-CXO联盟(cxounion.cn)
数字经济体现在产业数字化和数字产业化两个方面,产业数字化实际上就是把计算的能力、软件定义的能力、连接网络的能力等,嵌入到生产活动中去,而数字产业化则是把支持这些能力的数字技术变成产品,今天我们看到的计算机产业、软件产业、网络产业、互联网产业都属于这个数字技术产业化的过程。
数字经济的发展需要用软件来定义、增值、赋能,需要通过技术连接来形成更广泛的网络效应,技术连接在一起形成算力。算力已成为数字经济发展的重要驱动力,就像早年的工业化时代,能源、电力作为驱动力一样,它具体的表现就是计算无处不在,而数字芯片是算力的主要物理载体。
人工智能对集中使用的算力有了更大、更迫切的需求,需要把分布的算力集中起来处理大规模数据。因此我们看到了很多算力中心的建立,按应用场景可以分为智算中心、超算中心等,这些算力都可以称为云,而更加靠近用户的算力叫边,无处不在的嵌入式算力叫端,云边端整体构成了完整的算力基础设施。未来算力也将像电一样,实现端到中心的供给布局。
二、人工智能的终极目标是让机器像人一样
Q:人类发展人工智能的终极目标是什么?
王怀民:在计算机出现之前,所有计算都是靠人工完成的,人们把计算作为高水平的智力体现,计算机发明的初衷就是帮助人类处理复杂的计算问题。
人工智能,就是用一个人工装置去模拟自由人的智力才能完成的事,没有现代计算机,无所谓人工智能,因为没有一个可以模拟人的智力的人工装置的存在,所以人工智能发展的终极目标是让机器越来越多地能够承担起人的智力才能做的事。 CXO UNION-CXO联盟(cxounion.cn)
当然,现在大家说人工智能发展的终极目标是通用人工智能,比如一个程序能够做一个计算,能不能让它什么计算都能做,以前人工智能只能识别车牌,是不是以后什么东西都能识别,人类在追求这个。
但实际上,更广泛的意义上讲,人们现在对于什么是智能都说不清楚,我们可以举例说,但并不完整,智能本身是一个在探索中不断丰富的过程,因此我们说的目标,也是基于当下的想象,让机器能够做人类智能能做的更多的事情,简单地说,就是像人一样。
三、人工智能不是让人走开 是让人和机器更好合作
Q:今年这波由生成式AI所引发的人工浪潮所激发的大量算力需求,与此前经济发展所需的算力有何不同?未来人工智能可能发展到怎样的程度,对人类生活工作产生怎样的实际变化或影响?
王怀民:是一次巨大的变化。
今年是中国第一台超级计算机银河机——银河一计算机诞生40周年,1983年它刚被鉴定,那时这台计算机超算1亿次,现在是我们普通手机的100/10、1%的算力都不到,因此算力是随着时间发展的。
今天用在人工智能里进行大模型训练的大算力,是我们集中使用算力的最高状态,这就像40年前,我们用银河机去计算一些特定领域的工程数值,相对于那个时候的小型计算机而言,它的力量非常大,因此它是一个相对概念。
今天人工智能的水平带来巨大飞跃的算力呈现形态,它还不是未来最终形态,但是它的确带来了巨大惊喜,就是用这样的算力来训练如此大规模的数据,能够生成超出每个人预期的大模型,大模型未来会带来更多的变化,它的发展,充满想象力,这也是它如此活跃的原因。 CXO UNION-CXO联盟(cxounion.cn)
大模型将带来哪些变化?比如未来的采访,跟大模型聊就可以,然后它就可以整理出一篇稿子,当然还需要人确认,因为这个世界是要有人负责的,大模型不负责。
如果你是一个没有判断力的编辑,它就代替你了,你觉得他不行就说明,你有你的判断力、价值观和标准,你可能会让他反复改,然后你改的稿子又会存储在数字空间里面,这个数字空间的数据可能又会被大模型下一次学习。
因此,人工智能是一个人机协作、持续学习、成长演化的新的数字经济生态,它不是让人走开,而是让人和机器更好地合作。我们的进步以后不再是我们自己个人读书了,也不再是我们一般意义下的个人的实践了,而是和一个无处不在的、甚至你都没有感知的人工智能系统,包括大模型的支持相伴,然后你的能力不断提高,效率也大幅提高了。
四、生态营造至关重要 中国有独特优势
Q:站在专家的角度,您认为我国要抢抓人工智能发展机遇,在算力方面,还需做出哪些努力?
王怀民:计算系统的硬件、软件,乃至支持算力应用和投送的算力网的构建,这些都需要发展,还有一个重要的差距是生态。
生态首先是由相互联系的人构成的,然后由相互联系的人所创作出来的,比如说模型、数据、应用的相互联系构成的,又和由此相联系的、产生的市场和市场要素相互联系和构成,所以在这方面,中国要高度关注,实际上中国在生态营造上有一个独特的优势,就是中国有庞大的消费市场和工程师群体,因此我国去推动一个新兴市场的构建的潜力巨大。
一些发达国家的技术,如果没有市场和消费者,它就无法成功,中国有着巨大的消费群体,因此中国这个市场和人力资源投向谁,投向或拥抱哪个技术,谁就会成功。所以中国的消费者要把自己看成是一股巨大的力量,有没有定力去拥抱最有价值、最有前途、最可持续性的技术。 CXO UNION-CXO联盟(cxounion.cn)
现在由于美国芯片禁令的影响,我国在发展人工智能方面无法获得国际先进芯片,而这恰恰给了中国发展芯片、生态等的机会,我国要把握好这个机遇,利用当下国际竞争缠斗的格局和敏感时节,把自己的能力塑造起来。一方面,大家抱在一起,完成对于中国市场和中国消费者的吸引,完成算力发展生态的营造,另一方面也要在这个机遇期内,不断地推出先进技术,使得我国不至于比对方的技术有明显的缺陷。
五、人工智能发展也需扬善除恶
Q:新技术从出现到大规模使用,需要经历一个相对较长的“孵化期”,对于算力技术来说,在成为基础设施大规模使用的过程中,会面临什么样的阻碍?我国同步还需要做好哪些配套工作,比如算力安全、法律监管,您有何建议?
王怀民:生态的培育是一个长期的过程,这个过程中,关键是我们能够锁住和我们同行的消费者和开发者,当然资金环境也是生态的一部分,在这方面,需要有更好的支持产业发展的政策,更好地引导消费者使用的一系列政策。
在全球化背景下,我们一定要以开放、包容的竞争姿态,通过包括市场、政府、社会等在内的一种力量共同发力、营造生态。
另外,技术是造福人类的,技术本身是中性的,没有善和恶,但如果新兴技术被滥用,就可能生恶,因此人工智能、大模型的发展也需要扬善除恶。扬的是人的善,除的是人的恶,人的善恶在人类文明历史上早就有,人类已经探索出一系列扬善除恶的有效的社会机制,我们把它们用好就可以。
比如大模型生成的作品必须声明,必须有一个使用大模型产生这些作品的使用者的署名,出了问题这个人负责,这样就可以有效地推动大模型健康发展。 CXO UNION-CXO联盟(cxounion.cn)

翻译:
Wang Huaimin: The development of artificial intelligence is to make people and machines cooperate better
Under the development wave of artificial intelligence and digital economy, computing power is everywhere. What is the relationship between them? What is the ultimate goal of AI development? In order to seize the opportunity of artificial intelligence, what aspects of our country need to do?
With these questions, the reporter talked to Wang Huaimin, academician of the Chinese Academy of Sciences and member of the China Digital Economy 50 Forum, to understand the industrial logic and future trend behind the development wave of artificial intelligence and computing power.
Wang Huaimin (left)
Wang Huaimin, Academician of Chinese Academy of Sciences, member of Chinese Computer Society (CCF), Director of CCF Open Source Development Committee, National Leader in the Millions of Talents Project, member of the Science and Technology Commission of the Ministry of Education, has been engaged in research and talent training in the field of distributed computing for a long time, and has made systematic and creative work in distributed computing models, technologies and platforms. For our country’s distributed computing core key technology independent innovation, for the national and military network information system construction and development has made outstanding contributions. As the first complete person, Academician Wang Huaimin has won the National Technological Invention Award and the National Technological Progress Award many times.
Wang Huaimin said that artificial intelligence is a human-computer collaboration, continuous learning, growth and evolution of the new digital economic ecology, it is not to let people walk away, but to let people and machines better cooperation. CXO UNION-CXO联盟(cxounion.cn)
First, computing power is the driving force for the development of digital economy
Q: In the context of the current development of the digital economy, all walks of life are increasing computing power. How do you understand the relationship between computing power and digital economy and artificial intelligence?
Wang Huaimin: The digital economy cannot be separated from computing power, and it is precisely because of the development of computing technology that today’s digital economy is possible, so computing power is the technical basis for the formation of digital economy cognition.
Digital economy now has two aspects: industrial digitalization and digital industrialization. Industrial digitalization is actually to embed computing capabilities, software-defined capabilities, and network connectivity capabilities into production activities, while digital industrialization is to turn digital technologies that support these capabilities into products. The computer industry, software industry, network industry and Internet industry we see today all belong to this process of digital technology industrialization.
The development of the digital economy requires software to define, add value, and empower, and broader network effects need to be formed through technological connectivity, and technologies are connected together to form computing power. Computing power has become an important driving force for the development of the digital economy, just like in the early industrial era, energy and electricity as a driving force, its specific performance is that computing is everywhere, and digital chips are the main physical carrier of computing power. CXO UNION-CXO联盟(cxounion.cn)
Artificial intelligence has a greater and more urgent need for centralized computing power, and it needs to concentrate distributed computing power to process large-scale data. Therefore, we have seen the establishment of a lot of computing power center, according to the application scenario can be divided into intelligent computing center, supercomputing center, etc., these computing power can be called the cloud, and the computing power closer to the user is called the edge, the ubiquitous embedded computing power is called the end, the cloud edge end as a whole constitutes a complete computing power infrastructure. In the future, computing power will also be like electricity, achieving an end-to-end supply layout.
Second, the ultimate goal of artificial intelligence is to make machines like humans
Q: What is the ultimate goal of human development of AI?
Wang Huaimin: Before the emergence of computers, all calculations were done by humans, and people regarded computing as the embodiment of a high level of intelligence. The original intention of computer invention was to help humans deal with complex computing problems.
Artificial intelligence, is to use an artificial device to simulate the intelligence of free people to complete things, there is no modern computer, there is no artificial intelligence, because there is no artificial device that can simulate human intelligence, so the ultimate goal of the development of artificial intelligence is to let the machine more and more can bear the intelligence of people to do things. CXO UNION-CXO联盟(cxounion.cn)
Of course, now we say that the ultimate goal of the development of artificial intelligence is general artificial intelligence, such as a program can do a calculation, can it make any calculation can do, artificial intelligence can only identify license plates before, is it not possible to identify anything later, human beings are pursuing this.
But in fact, in a broader sense, people are not clear about what intelligence is now, we can give examples, but not complete, intelligence itself is a process of continuous enrichment in exploration, so the goal we say is also based on the imagination of the moment, so that machines can do more things that human intelligence can do, simply put, like people.
Third, artificial intelligence is not to let people walk away is to let people and machines better cooperation
Q: How is the massive demand for computing power generated by this year’s wave of artificial labor generated by generative AI different from the computing power required for previous economic development? To what extent is artificial intelligence likely to develop in the future, and what actual changes or impacts will it have on human life and work?
Wang: It’s a huge change.
This year is the 40th anniversary of the birth of China’s first supercomputer Galaxy machine – Galaxy a computer, it was just identified in 1983, when the computer supercalculated 100 million times, now is our ordinary mobile phone 100/10, 1% of the computing power is less than, so the computing power is developed with time. CXO UNION-CXO联盟(cxounion.cn)
Today, the large computing power used in large-scale model training in artificial intelligence is the highest state of our concentrated use of computing power, which is just like 40 years ago, we used the Galaxy machine to calculate some specific fields of engineering values, relative to the small computer at that time, its power is very large, so it is a relative concept.
Today’s level of artificial intelligence has brought a huge leap in the form of computing power, it is not the final form of the future, but it does bring a huge surprise, that is, to use such computing power to train such large-scale data, to generate large models beyond everyone’s expectations, large models will bring more changes in the future, its development, full of imagination, which is why it is so active.
What changes will the big model bring? For example, in the future interview, you can talk to the big model, and then it can sort out an article, of course, it needs to be confirmed, because the world is responsible for someone, the big model is not responsible.
If you are an editor without judgment, it will replace you, you think he can’t show that you have your judgment, values and standards, you may let him change repeatedly, and then your changed manuscript will be stored in the digital space, the data of this digital space may be learned by the big model next time.
Therefore, artificial intelligence is a human-computer collaboration, continuous learning, growth and evolution of the new digital economic ecology, it is not to let people walk away, but to let people and machines better cooperation. Our progress is no longer our own personal reading, nor is it our general sense of personal practice, but with a ubiquitous artificial intelligence system that you do not even perceive, including the support of large models, and then your ability continues to improve, and efficiency is greatly improved. CXO UNION-CXO联盟(cxounion.cn)
Fourth, ecological Construction is crucial China has unique advantages
Q: From the perspective of experts, what efforts do you think China needs to make in terms of computing power in order to seize the development opportunity of artificial intelligence?
Wang Huaimin: The hardware and software of the computing system, and even the construction of the computing network supporting the application and delivery of computing power, all need to be developed, and there is an important gap in ecology.
Ecology is first composed of interconnected people and then created by interconnected people, such as the interconnection of models, data and applications, and the interconnection and interconnection of the resulting market and market elements, so China should pay great attention to this aspect. In fact, China has a unique advantage in ecological construction. China has a huge consumer market and a large group of engineers, so there is huge potential for China to promote the construction of an emerging market.
Some developed countries technology, if there is no market and consumers, it can not succeed, China has a huge consumer group, so China’s market and human resources to whom, to which technology or embrace, who will be successful. So Chinese consumers need to see themselves as a huge force, with no determination to embrace the most valuable, promising, and sustainable technologies.
Now due to the impact of the United States chip ban, China cannot obtain international advanced chips in the development of artificial intelligence, and this precisely gives China the opportunity to develop chips, ecology, etc., China should grasp this opportunity, and use the current pattern of international competition and sensitive season to shape up their own capabilities. On the one hand, we hold together to complete the attraction of the Chinese market and Chinese consumers, and complete the construction of the development ecology of computing power, on the other hand, we must continue to introduce advanced technologies in this period of opportunity, so that China will not have obvious defects than each other’s technology. CXO UNION-CXO联盟(cxounion.cn)
Fifth, the development of artificial intelligence also needs to promote good and eliminate evil
Q: New technologies need to go through a relatively long incubation period from their emergence to large-scale use. For computing technologies, what obstacles will they face in the process of becoming large-scale use of infrastructure? What supporting work needs to be done in China’s synchronization, such as computer security and legal supervision, do you have any suggestions?
Wang Huaimin: The cultivation of ecology is a long-term process, in this process, the key is that we can lock in consumers and developers with our peers, of course, the financial environment is also a part of the ecology, in this regard, we need to have better policies to support industrial development, and better guide consumers to use a series of policies.
In the context of globalization, we must adopt an open and inclusive attitude of competition, and make joint efforts through the market, the government and society to create an ecological environment.
In addition, technology is beneficial to mankind, technology itself is neutral, there is no good and evil, but if the emerging technology is abused, it may produce evil, so the development of artificial intelligence and large models also need to promote good and eliminate evil. The good and evil of man have long existed in the history of human civilization, and mankind has explored a series of effective social mechanisms for promoting good and eliminating evil, so we can use them well.
For example, the works generated by the large model must be declared, and there must be a signature of the user who uses the large model to produce these works, and this person is responsible for the problem, so that the healthy development of the large model can be effectively promoted.
由CXO UNION-CXO联盟(cxounion.cn)转载而成,来源于证券时报;编辑/翻译:CXO UNIONCXO联盟小U。
如需加入CXO UNION(CXO联盟)高管社群,请联系社群小伙伴哦~

免责声明: 本网站(http://www.cxounion.cn/)内容主要来自原创、合作媒体供稿和第三方投稿,凡在本网站出现的信息,均仅供参考。本网站将尽力确保所提供信息的准确性及可靠性,但不保证有关资料的准确性及可靠性,读者在使用前请进一步核实,并对任何自主决定的行为负责。本网站对有关资料所引致的错误、不确或遗漏,概不负任何法律责任。
本网站刊载的所有内容(包括但不仅限文字、图片、LOGO、音频、视频、软件、程序等) 版权归原作者所有。任何单位或个人认为本网站中的内容可能涉嫌侵犯其知识产权或存在不实内容时,请及时通知本站,予以删除。
如需加入CXO UNION(CXO联盟)高管社群,请联系社群小伙伴哦~

免责声明: 本网站(http://www.cxounion.cn/)内容主要来自原创、合作媒体供稿和第三方投稿,凡在本网站出现的信息,均仅供参考。本网站将尽力确保所提供信息的准确性及可靠性,但不保证有关资料的准确性及可靠性,读者在使用前请进一步核实,并对任何自主决定的行为负责。本网站对有关资料所引致的错误、不确或遗漏,概不负任何法律责任。
本网站刊载的所有内容(包括但不仅限文字、图片、LOGO、音频、视频、软件、程序等) 版权归原作者所有。任何单位或个人认为本网站中的内容可能涉嫌侵犯其知识产权或存在不实内容时,请及时通知本站,予以删除。
Search
Popular Posts
-
2024数字化灯塔案例评选申报开启!
“2024数字化灯塔案例评选”于3月正式启动,诚挚欢迎业界同仁自荐和推荐,一起推动产业数字化进程,助力赋能企业…
-
2024 X-Award星盘奖申报通道已开启!
X-Award星盘奖是数字化转型服务、IT服务行业重要的商业奖项,旨在表彰行业里提供杰出数字化转型服务与IT服…
-
2024 N-Award星云奖申报通道已开启!
N-Award是数字化转型领域重要的商业奖项,旨在表彰那些以非凡的远见、超群的领导才能和卓越的成就来激励他人的…




