Set micro-network launched integrated circuit micro-letter public number: ' Every day IC ', major news real-time release, every day IC, every day set micro-nets, accumulate micro-Cheng! Replication Laoyaoic Micro-letter public number search adds attention.
1. China technology companies to try more new areas of AI;
Compared to Silicon Valley, China Baidu, Alibaba and Tencent have more data advantages in AI. Chinese companies have used machine learning to enter a number of newer areas, including medical diagnostics, facial recognition, and the use of voice-operated AI hardware.
The FT reported that Baidu, Alibaba and Tencent attached great importance to AI and established expert laboratories and hired top engineers at home and abroad. Tencent chief operating officer Ningyu said that in 2016 Tencent began to regard AI as one of the priorities, as a strategic focus.
Alibaba in this year's earnings report also began to introduce the results of the promotion of AI. November 11 this year, "Singles Day" shows a part of the power of AI, Alibaba in the shopping festival one months before the use of AI produced 400 million custom banner ads; In the pre-sale period, Alibaba's chat robot answers 3.5 million simple questions every day, such as "Where is my parcel?" 」
Alibaba's third party electric business platform Taobao is also developing AI, allowing users to use the image search, to find the same or similar products. For example, this allows shoppers to search for a dress worn by a celebrity.
Analysts say the Chinese companies are still in the early stages of using AI, but such applications suggest how China could be a global leader, especially as companies speed up the use of relevant technologies.
Anand Swaminathan, a senior partner at McKinsey, a consultancy, Svaminaschan that China has two advantages over Silicon Valley: "They are here to test more than 1 billion people, so the US is congenitally deficient." And China's investment rate and scale are fundamentally higher than the US. 」
How the Chinese science and technology industry may use AI in the communications and healthcare sectors will be a topic of concern in the future.
Alibaba currently employs the voice of the concept of personal assistants, than Apple's Siri and Amazon's Alexa one step ahead; Alibaba plans to introduce a ticket machine that filters out background noise and listens only to ticket-makers at subway stations.
Zhihijie, director of Alibaba's intelligent voice interaction team, said the technology was novel in that it was applied to actual products with the help of improved technology, and that personal assistants could enter public places, including airports, railway stations, restaurants and desks.
More intelligent AI and larger data scales also apply to medical applications. Tencent intends to apply AI technology to one of the companies that detects early lung cancer. But Lin Zhilang, a professor of respiratory medicine at the University of Hong Kong, said it would take 5 years for the project to arrive at preliminary results. Central Society
2. Baidu, Huawei AI strategic cooperation and intelligent terminal soft and hard fusion;
Baidu, Huawei announced 21st in Beijing signed a strategic agreement between the two sides in the Internet services, content ecology and artificial intelligence AI field of in-depth cooperation. including Baidu Voice, visual recognition and other artificial intelligence technology are placed in Huawei mobile phones; In addition, Baidu's content ecology will also and Huawei mobile phone deep cooperation. This cooperation, Baidu will obtain artificial intelligence terminal commercialization important way; And Huawei can use Baidu's technology platform to enhance the mobile phone user experience, both sides. Baidu CEO Robin Li said Baidu and Huawei have very similar genes. But the pursuit of technology is a bigger common denominator. Now Baidu and Huawei holding hands, is ' flowers (hua) qi put '. Huawei Yu said, in the Huawei mobile phone, the input method is now used in Baidu, search is using Baidu, the future will use more Baidu's business, including Baidu Voice, Baidu map. In the future, both sides will cooperate in the field of AI. Yu said that Huawei has just released the mobile phone mate10 has been implanted AI chip, the future will continue to strengthen. Yu further pointed out that the mobile phone industry highly competitive in the future to survive only a few companies, the future hope to do more than 40% of the city accounted for. Huawei needs Eco-partners to build the ecosystem of hardware and software. Yu said he observed many internet companies and found that Huawei and Baidu were the most highly complementary. It is reported that the details of cooperation between the two parties include: natural interaction; Paddlepaddle+hiai (including Kirin 970 chip and paddlepaddle cooperation); Content and services. The two sides said that the company will deeply integrate Huawei's mobile communications and intelligent terminal products, and Baidu's advantages in AI capabilities, content capabilities, and business capabilities to jointly explore the infinite possibilities of the age of wisdom. In the AI platform and technical cooperation, the two sides will be based on Huawei Hiai platform and Baidu Paddlepaddle in-depth learning architecture, to play the Huawei Npu (neural receptacle 處理 unit) The advantages of To build an open environment with Huawei AI open platform and Baidu Brain as the center, and to serve the vast number of AI developers to enrich the consumer AI applications and intelligent service experience. At the same time, both sides will be in all kinds of intelligent terminals to carry out in-depth voice and image cooperation, promote human-machine communication more convenient; For the AR scene innovation, the two sides will be technical innovation + hardware landing of the soft and hard combined form to create AR ecology, for the vast number of users to open up the boundaries between virtual and reality; In the area of Internet service and content ecological cooperation, both sides will strengthen cooperation in the fields of search and information flow, and jointly bring richer quality content and intelligent and convenient service experience to consumers. In short, the cooperation between the two sides includes: 1. Natural interactions, including voice, semantic, visual and immersive experiences; 2.paddlepaddle+hiai (including Kirin 970 chip and paddlepaddle cooperation); 3. Content and services, including search, information flow, mobile phone Baidu. In fact, Baidu in the AI field of recent moves. November, Huawei and Baidu know to reach strategic cooperation, through product knowledge services and cloud technology, NLP and other AI depth learning capabilities. Then December, Qualcomm announced and Baidu Dueros cooperation, focus on Yu Juanlong platform mobile phone intelligent voice system; En NXP also announced, officially become Baidu Apollo Open platform partner. Just one months ago, at the Millet Networking developer Conference, Lurkey and Lei, Baidu and Millet jointly announced that will be in the field of things networking and artificial intelligence cooperation. Digitimes
3.AI Evolution to promote the new industrial revolution;
The generation of artificial intelligence (AI), is by many different fields of science and technology integration, a single technology advance without relevant advanced hardware and software auxiliary, it is difficult to promote the AI to a new realm.
In the near future, on the shoulders of Giants, a group of catalysis AI emerging technologies have been assembled, including the neural network, such as depth learning algorithm breakthrough, the combination of large data and cloud computing, the continuous evolution of operational efficiency under Moore's Law, the rise of all kinds of sensory systems, in all things "can connect" and "speed Company" platform, To ignite the dormant AI accelerator. Let AI with updated thinking (calculus architecture), brain tissue (cloud hardware), supplemented by internal higher-speed computing, and larger amounts of data (large data) are used to reveal human intelligence through self-learning and thinking (deep learning) after a variety of external messages (IoT sensors). Ai in the way humans think, with machines are entering the real life of the people.
The current "weak AI" such as voice assistants, image recognition, intelligent translation and sweeping robots, and so on, do help people solve many problems, but also make life more convenient and efficient, these digital forces as human beings, to help share the day-to-day needs of mental and labor; And the "strength" of the AI, such as self-driving, industrial robots, intelligent financial advisers, or even artificial intelligence physicians or lawyers, the extent of sharing and the ability to solve problems, more generally, such as self-driving car accident rate can be reduced, industrial robots can improve production efficiency, AI Consultants can provide more rational and higher profit probability recommendations, intelligent medical care can enhance the health care resources and quality, so that people really enjoy life.
Especially for stability, easy to predict and replicate the work, AI can even do faster and better than humans, which means that no matter whether the human work has been robbed, in the future in various AI popularization, our working patterns will certainly change, this change is not limited to a single domain or industry, It is a paradigm shift that covers a variety of professions. Through learning, AI is quietly driving the next generation of the industrial revolution.
In the previous industrial revolutions, machines may have simply replicated human brains or labor, but now machines or programs have learned to replicate the "profession" of humanity. For example, the financial industry used the machine to help the Commissioner to do analysis and advice to customers, the machine itself only with storage, operation, data collection and validation, and other ancillary functions, but the future financial industry can use the professional ability of the intelligent machine, direct analysis and decision-making.
The future, such as the concept of "duplicating a profession" for machines (artificial intelligence), may be as simple as configuring machine modules; Some may say that it is for the simple professional, but the nature of artificial intelligence is to continue to learn and grow, now the so-called simple professional, for the past is extremely difficult. After the relevant technology is in place, believe that for all types of "professional" replication will be the next development trend of artificial intelligence, and its impact level not only changes our working style, and thus derived from the education, society, crime, or even the legal form of change, Ai unlimited potential is to make human more efficient pursuit of desire, But is the cost and risk of relative change really what we can afford?
Whether we think of AI as hope or doom and gloom, the rotating technology is never stopped because of human 迟豫, electronic communication will not return to the Flying Pigeon book, the electric car train will not return to prancing whip, holding only time to think about coping with this impending smart storm should be more practical. If there is no essential difference between a human brain and a computer, as the scientist Hawking says, the most feared risk is not how the machine can quickly learn the human race. But as Apple's CEO Cook put it, humans are becoming more of a computer-like thinker, increasingly filled with worthless, empathetic and reckless ideas, like machine thinking, which are gradually assimilated by numbers. (the author is researcher of Science and technology Policy Research and Information Center of Taiwan National Experimental Institute)
Economic Daily
4. Anti-AI technology abuse Europe and the United States to build firewalls;
With the current technology of intelligent technology, film or novel "Strong Artificial Intelligence" situation is still a distance from the actual, hidden behind the potential risks to the general people feel more remote as the sky, and the human habit of the future is often after the occurrence of the discussion, the threat of artificial intelligence has been only out of everyone's mouth.
In fact, the world has begun to build artificial intelligence risk of the firewall, from various levels to examine the future of the various conditions, early to do risk prevention and control, its content covers policy, science and technology, law, ethics and is a survival issues.
On February 16, 2017, the European Parliament formally proposed a bill to the European Commission to set up a comprehensive law to define the moral and legal liability of artificial intelligence, and the legislative motive lies in the rapid development of artificial intelligence and robotics, which gradually becomes part of human daily and economic activities, It is hoped that the legislative norm will ensure its safety. At present, the legislative resolution is still in the discussion stage, the future will set up specialized agencies to implement the relevant artificial intelligence and robotics in the technical and ethical supervision.
The White House president's office also published in December 2016 the AI, the white Paper on automation and economics illustrates the challenges that the labour market, economic and policy decrees may face in the future under the impact of artificial intelligence, and proposes new education, training and transition strategies that are needed to change the future working patterns. The British science Office also released its policy report on November 9, 2016, "Artificial Intelligence: Opportunities and future decisions," introducing the opportunities and impacts of artificial intelligence on society and Government, and proposing the responsibility of adapting to artificial intelligence for decision making and risk management mechanisms.
In industry, Google in 2014 proposed the establishment of an ethics committee to ensure that artificial intelligence robot technology is not abused. OpenAI was founded in 2015 by Tesla, Y Combinator and other companies, focusing mainly on the ethical issues of artificial intelligence. In addition, Facebook, Amazon, Alphabet, IBM and Microsoft, the five major technology industry in 2016 announced the establishment of "the benefit of the crowd and social Artificial Intelligence Partnership" (Partnership on AI), which is a non-profit organization, hoping to pass a standard open model, In the process of social change, artificial intelligence becomes a research subject which includes morals, fairness, inclusiveness, transparency, privacy, interaction, human and artificial intelligence, trust and reliability. It is hoped that this kind of coverage of human core issues will be developed to dispel the suspicion of artificial intelligence risks.
The prevention and control of the learning and research circles is vigorous and diverse, such as the Standard Development Organization IEEE published "Ethical Design: the use of artificial intelligence and autonomous systems to maximize human well-being of the vision" is to encourage scientific and technological personnel in the AI research and development process, the priority to consider ethical issues. January 2017 beneficial AI 2017 conference held in California, the world's top AI field experts jointly signed the 23 principles of future AI development (Asilomar AI Principles), including the security of AI development, Responsibility and respond to the catastrophic risk posed by AI. The Center for Risk Research, an interdisciplinary research center within the University of Cambridge, is dedicated to studying and mitigating the risks that could lead to human extinction or a collapse of civilization, which lists AI as one of the major risks that could cause human extinction, in addition to warning, and research and discussion of prevention methods. (the author is researcher of Science and technology Policy Research and Information Center of Taiwan National Experimental Institute)
STPI Introduction
The National Institute of Experimental Science and technology Policy Research and Information Center (STPI) was founded in 1974, has long been dedicated to China's scientific and technological research and development needs of data collection, construction, analysis, processing and services and other matters. In recent years, on the basis of existing information resources, strengthening trend analysis, key issues excavation, patent information analysis, Innovation and entrepreneurship promotion and other energy, and help the government to mark the development of China's science and technology vision and strategy, to the professional science and technology Policy Research think tank forward.
Trend analysis and forward planning team
This team takes the systematic method to carry on the technology development tendency observation and the thorough research analysis, provides our country to study the dynamic intelligence information to grasp the trend pulsation; and establish the evidence based on the demand-oriented forward-looking research energy, to assist the departments will carry out the direction of China's scientific and technological development planning.
Economic Daily
5.AI five big worries to beware;
People's risk of independent intelligence is expected to focus on the more situational imagination of "strong artificial intelligence", such as a highly intelligent robot; With the advent of iot/ai generation, "weak artificial intelligence" has infiltrated into human life, all sectors have to look at all the possible impact of artificial intelligence, and the risk.
The distribution of resources is extremely uneven: artificial intelligence has begun to change the working patterns of many people, and the 2016 Global Economic Forum predicts that artificial intelligence and machine learning will result in the disappearance of about 7 million jobs over the next few years, as well as the creation of 2 million new job opportunities. British physicist Stephen Hawking argues that the rise of artificial intelligence could hit the middle class and leave a handful of specific jobs, leading to serious economic imbalances.
These inferences suggest that human "professional" will be easily replicated by machines, resulting in structural adjustment of economic strata and the formation of technical unemployment, production benefits and profits will be concentrated in a few highly intelligent resource groups, rich and poor difference between the polar drama, social height exposure to the risk of uneven allocation of resources.
Information monopoly: Human thinking can not jump off the scope of their own experience, the machine is the same. Artificial intelligence without data, even with excellent operation of the hardware and algorithm, still can not play a role in the field of the target. In the future in the relevant software and hardware technology in place, "enough and useful (massive and tagged) data", will be the artificial intelligence industry to build a niche of one of the biggest key, each domain leader will be a large number of key data barriers to create a partition, and make the user access to data costs more and more high, resulting in "the wisdom of the wise, The Fool more Foolish "information monopoly risk."
No privacy in the future: McKinsey estimates, the 2025 will have 1 trillion kinds of objects interconnected, which means that in the trend of the Internet of things, may be all around the future of information collectors, under the artificial intelligence technology, these information collectors may have autonomic thinking function, and determine the time needed to open the sensor. Regardless of the future of the protection of the right to privacy can be achieved, artificial intelligence and the full effect of networking, the beginning of the risk of privacy to advance to a very high level.
Social alienation: In the network era prevailing, rely on the network virtual world to meet the social interaction of the number significantly improved, the network virtual world of friendliness, so that people with the exchange of imagination and opportunities, network virtual personality seems to have more courage and creativity; But this is not the real world, it will make the extremely dependent groups of the real world between the big gap, and thus cause social alienation. In the era of artificial intelligence, the virtual world has been extended, the virtual reality and artificial intelligence to make the virtual world more real, the emergence of more dependent clustering, resulting in the real social interaction more alienated phenomenon.
No autonomy: Artificial intelligence has evolved to judge the information we often see and want to see in our daily lives, and the machine-giving method helps save time, but it also single the message. In a world where AI objects are more prevalent in the future, this phenomenon of human over-reliance and being single value (or target value) will escalate in a more deteriorating direction, leading to the absence of independent thinking in all areas of human risk and, to a certain extent, the crisis of human mental extinction. (author is researcher of Science and technology Policy Research and Information Center of Taiwan National Experimental Institute) economic Daily
6. Chip into the AI ERA in addition to spell the technology to spell the scene
Ai chip in the field of war more and more intense, in addition to the technical level of breakthroughs, but also to fight who can seize more of the application scene.
The first AI vision chip, which lasted two years, on the horizon of AI start-ups, was launched on December 20, starting with three categories of intelligent driving, smart city, and smart business. Horizon founder Yukei told the first financial reporter that next year there will be more commercial landing in the field of smart home.
In the AI chip field, in recent years, along with the horizon, the deep science technology, the Cambrian and so on the rise, or will change our country each year spends the huge foreign exchange import chip the situation.
Combination of soft and hard route
Horizon This release of the chip includes two, a name Journey (groceries) 1.0, for intelligent driving, another named the Rising Sun (Sunrise) 1.0, the main smart camera. The two chips used are the first generation of the horizon BPU architecture-Gaussian architecture.
Horizon Company Introduction, ' journey ' can simultaneously to pedestrians, motor vehicles, non-motorized vehicles, lane lines, traffic signs, lights and other targets for real-time detection and identification. The ' Sunrise ' can implement large-scale face detection, video structure and other applications in front.
Yukei was one of the principals of the Institute of in-depth study of Baidu, has been engaged in the software aspects of the work. However, when the horizon was founded in 2015, the team opted for a combination of soft and hard lines, the ' algorithm + chip ' business model, which is a more ' heavy ' route. For the team is a cross-border, but also need a lot of money to support the long cycle of research and development.
Artificial intelligence to the various scenes in society, application, only software is not enough. ' Yukei said at the 20th press conference. He believes that the combination of soft and hard is now more and more attention in the forefront direction.
' We all know Alphago defeated Shishi, but we don't know how much we pay for it. The human brain is 20 watts, and the computer used by the Alphago is 2000-kilowatt, we use 100,000 times times the power to defeat the Shishi, which means that the artificial intelligence to develop forward, the hardware is certainly an obstacle. Zhang Yi, a professor of computer science at Tsinghua University and an academician of CAs, said at the scene.
Next: Application Scenarios
From the application scene to drive the hardware design, currently the leading two giants are Nvidia and Intel. The former GPU (graphics processor) is widely used in the field of AI. The latter has made a lot of acquisitions and investments in the AI chip industry chain, including about 100 million dollars before the horizon.
In addition, Google, Tesla, Apple, Qualcomm, Huawei and other industry giants are developing AI chips. The US investment Research agency Morningstar has previously released a report referring to the alphabet (Google's parent) 's TPU, which could drive companies such as Amazon and Microsoft to launch more customized chips to become rivals for Intel, Nvidia and Xilinx in the AI chip market.
These companies are trying to produce the products that best suit their needs to reduce reliance on traditional chip giants. Like Tesla, for example, its autopilot hardware system is based on the NVIDIA graphics GPU, but will have less reliance on NVIDIA once it has an independent research and development chip capability.
Some new unicorns in the AI field are also adding horsepower. Just last month, the Cambrian incubation of the Chinese Academy of Sciences released a new generation of AI chips. According to the Cambrian, the application of this chip covers the fields of image recognition, security monitoring, intelligent driving, UAV, speech recognition, natural language processing and other applications.
Referring to the difference between competitors, Yukei said, the horizon of AI chip selection is a combination of scene depth model, on this basis, the company an important business model is ' turnkey solution ', that is, the provision of software and hardware design with a customized solution.
Google Cloud Artificial Intelligence and machine learning scientist Li Feifei said earlier, artificial intelligence is going through the historical moment, it has gone out of the laboratory into the industrial application phase.
For the horizon, the first vertical area to be preempted is autopilot, which they think is the ' Everest ' of the core technology of artificial intelligence. The horizon has worked with Audi to develop platform-driven self-driving technologies for the global marketplace. In addition, it is working with Intel to develop a new type of automated processor architecture, as well as a software system for vehicle specification (vehicle navigation standards).
In addition, the horizon is also a commercial attempt in the field of intelligent home, including intelligent air-conditioning, sweeping robots and other products. Yukei to the first financial reporter that next year in the second generation of Bernoulli architecture of the new generation of the processor launched, in the commercialization of smart home will still have more landing.
First Financial