In recent years, advances in data collection and integration, algorithms, and computer processing have led scientists and engineers to make significant advances in the development of artificial intelligence (AI). Suddenly, the machine has been able to accomplish tasks that once required human cognition. In the past, computers were only able to execute fixed programs that were already written. Now, people can provide a common strategy for computer learning that enables them to process new data without having to be reprogrammed. At present, many "machine learning" systems like this have been put into commercial use. In the fields of finance, healthcare and transportation, the "machine learning" system is becoming more and more widely used. These systems have begun to have an impact in ten ASEAN countries (ASEAN, English name ASEAN).

The two major global centers for artificial intelligence development are the United States and China. The United States has taken the lead in developing many applications, and China is still in second place, but its development momentum is rapid. In contrast, ASEAN countries are lagging behind, but each of them has some progress in artificial intelligence. Among them, Singapore has the most research results, and in Malaysia and Vietnam, it is very encouraging to see some early results. Although artificial intelligence tools are being adopted in industries such as transportation, financial services, healthcare, and media, the technology industry is still at the forefront of development.

Because artificial intelligence technology can significantly increase productivity, it can have devastating effects on the economy of Southeast Asia and the workers there. Previously published MGI research estimates that existing artificial intelligence technologies are likely to automate half of the existing work activities of the four ASEAN economies: Indonesia (52%), Malaysia (51%), the Philippines (48%) and Thailand ( 55%). These jobs currently generate more than $900 billion in wages.

However, it is worth noting that technical feasibility is not the only factor that affects job automation. Companies will also consider factors such as cost of purchasing and implementing technology, labor market dynamics, business interests, regulation, and social recognition. But it seems that the nature of many existing jobs will change, and more and more jobs need to interact with machines, for which Southeast Asia will need to develop new types of workforce skills.

If AI technology is used in the right way, it is likely to have a positive social impact for ASEAN countries. For example, machine learning innovation can improve credit models and enhance financial inclusion. Artificial intelligence solutions can provide new types of preventive and telemedicine, assist in disease diagnosis, and accelerate the development of new drugs. Adaptive learning algorithms can play a role in virtual education and personalized teaching. To achieve these uses, most parts of Southeast Asia will need to establish basic digital facilities and data ecosystems.

For most ASEAN companies, they will need to make fundamental changes to their management culture, including data-driven decision making; the most important of these is to build innovative partnerships with professional companies to incubate artificial intelligence. Scarce skills required in the field. In addition, companies need to prioritize funding and manpower to develop effective ways to strengthen data infrastructure.

Although the market will become a driving force for the development and popularization of artificial intelligence, the government still needs to play a key role in bringing benefits to society as a whole. In summary, there are three key points: establishing a regional policy framework to support the development and popularization of artificial intelligence; developing artificial intelligence talents and encouraging use at the local level; focusing public opinion on artificial intelligence to achieve inclusive growth, Bring positive social impact.

Figure: Artificial intelligence global investment is growing rapidly

Figure: Artificial intelligence technology brings many opportunities to Southeast Asia

The potential social benefits of artificial intelligence technology after large-scale application: The first is that machine learning innovation can improve the credit model and enhance financial inclusiveness; the second is that artificial intelligence solutions can provide new preventive medical and telemedicine, and assist Disease diagnosis can also speed up the development of new drugs; thirdly, adaptive learning algorithms can play a role in virtual education and personalized teaching.

Figure: Five basic elements of using artificial intelligence within the company

Artificial intelligence (AI) refers to the ability of machines to display human intelligence. For example, the ability to solve problems without the need for detailed, manually developed software assistance. By consulting a large number of pattern datasets, machines can "learn" to perform tasks such as identifying images, recognizing speech, identifying relevant information in text, integrating information, drawing conclusions, and making predictions. With the rapid development of artificial intelligence technology, its practicality has also been improved in more and more fields.

There is no universally consistent definition of the composition of artificial intelligence. This area is currently evolving rapidly, and developers often combine existing technologies to solve specific problems. Therefore, the term "artificial intelligence" covers a wide range of technologies and applications, some of which are extensions of earlier technologies (such as machine learning), and others are other entirely new technologies. In fact, there is no “smart” theory that is generally accepted, and the definition of artificial intelligence is constantly changing as people's understanding of current progress. Although there is disagreement about how to define boundaries in this area, one thing is widely shared: artificial intelligence will bring the next wave of digital subversion.

Artificial intelligence and the employment opportunities it brings

People are increasingly aware of (and anxious about) that artificial intelligence can have devastating effects on the labor market. The previously released MGI research project shows that almost half of the current work activities are automated, which can be achieved by adapting existing technologies. Currently, 30% of activities in 60% of occupations can be achieved through automated technology (see Figure 1). Since automated applications are at the task level, artificial intelligence seems to change more and more occupations, but it will not be completely eliminated. This may have a profound impact on the demand for certain workforce skills in Southeast Asia and may exacerbate labor market turmoil.

Exhibit 1: Although there are few occupations that can be fully automated, 60% of current occupations include at least 30% of activities that can be done through automated technology. 5% of occupations can be 100% automated, and approximately 60% of occupations include at least 30% of work that can be done automatically.

Occupations that contain a large amount of data collection, processing, or fixed procedures in their work will be affected first. These occupations include food processing, office management and factory production. This impact on the labor market can be enormous. In Southeast Asia, MGI found that existing technologies can automate more than half of the work activities in Indonesia, Malaysia, the Philippines and Thailand, with 52%, 51%, 48% and 55% respectively. These jobs currently generate more than $900 billion in wages. But this does not mean that, just because it is technically feasible, companies will replace workers with machines overnight. The speed and scope of enterprise automation will depend on how they view business cases, weigh the cost and ease of use of these technology systems, labor market dynamics, possible value creation, customer experience, capabilities, regulation, and social acceptance.

Artificial intelligence has the potential to dramatically increase productivity due to technology disruptions that have occurred in the past, and productivity has always been the key to generating revenue growth and promoting economic prosperity. According to MGI estimates, assuming that the replaced human workforce can be redeployed to jobs that are still as efficient as in 2014, the spread of artificial intelligence will increase global GDP by 0.8% to 1.4% annually. But this opportunity not only increases efficiency today, it also creates new avenues for future growth. In our survey of “artificial intelligence” executives, 20% of them used labor cost savings as their primary motivation for artificial intelligence. But more people (25%) said they introduced artificial intelligence to expand their business.

Artificial intelligence technology is approaching a tipping point

People's expectations for artificial intelligence are soaring. As costs fall, the computing power of computers is growing exponentially, enabling companies and organizations to run complex algorithms on large data sets. Today, algorithms that operate on large data sets have transcended human capabilities in areas such as image and speech recognition. Perhaps most importantly, with the spread of mobile devices, a large amount of new data is being generated that reflects all aspects of the consumer's lifestyle and how it is consumed. In fact, the world generates about 2.2 EB (2.2 billion GB) of new data every day. These trends have led to significant advances in the field of artificial intelligence – and these advances have moved out of the laboratory into real business applications (see Exhibit 2).

Machine learning is the basis of most current artificial intelligence use cases. It is based on algorithms that update by identifying patterns in large data sets without the need for rule-based programming to operate or draw conclusions. In practice, machine learning can be used to study project outcomes such as consumer demand or public health needs. It also optimizes equipment maintenance, adjusts prices, adjusts marketing information, and delivers a personalized retail experience.

Figure 2: Due to the emergence of various favorable conditions, the popularity of artificial intelligence is reaching a critical point. Connected devices and data availability (in billions), improved algorithms and training methods (error rate, %).

In addition: combined with information processing (such as computer vision and natural language) and drive technology (robots and autonomous vehicles), machine learning technology has the potential to change many aspects of our daily lives (see Figure 3).

Exhibit 3: Artificial intelligence will change our daily habits to improve quality of life and human productivity.

TuTI is a mother whose daughter is Andira. She is also a middle manager of a green energy company based in Jakarta.

(1) In the morning, her virtual assistant helped her to preview the plan for the day, including: the doctor gave her a health reminder. Confirm the order for the groceries to be delivered that evening. Andira's school sent notifications about homework and how TuTI interacted with her based on machine learning.

(2) TuTI and Andira read together on the road to the school by driverless car, which is safer, smoother and faster than driving, thanks to intelligent route planning and traffic monitoring.

(3) At the company, TuTI is the chief engineer of a wind farm. Her virtual assistant provided her with the latest information in this field. It helps Tuti sort out the tasks that need to be prioritized that day, because it already knows when and how Tuti works most efficiently. The system's output mode is abnormal, drawing her attention to one of the generators; but the robot can remotely inspect and repair it, minimizing production loss and safety hazards.

(4) After work, Tuti went to the gym for a quick workout. The virtual assistant refers to Tuti's health notice and proposes changes to her workout schedule.

(5) Back home, Tuti helps Andira to complete homework, and there is a virtual tutor next to it. The tutor's content comes from a virtual teacher who also collects data during the process of guiding Andira and incorporates it into an anonymous country. Level database to improve the overall quality of teaching in Indonesia.

Although the pace of change is uncertain, the changes are indeed happening and have begun to penetrate traditional non-technical industries such as manufacturing. For example, Foxconn deployed 40,000 robots in its Chinese factories, making the factory less labor intensive. In fact, Foxconn is still planning further, with the goal of producing 10,000 robots within the company to achieve its industrial automation goals.

Declining costs, modularity of technology components, and the development of user-friendly tools and interfaces are rapidly making artificial intelligence a viable and necessary operational asset for more and more companies. They can now connect off-the-shelf data management platforms to their most important assets to get the lowest upfront costs.

Early adopters of artificial intelligence will take advantage

As the early wave of digital technologies showed, early adopters of artificial intelligence can take a major competitive advantage and continue to maintain this advantage over time—especially if they see this new technology as a Key business capabilities and future sources of revenue growth, not just a means of cutting costs. Big companies do better in this area because they have the ability to invest in early trials and achieve greater return on investment by gradually expanding the scale of their business.

Our estimates show that in 2016, companies worldwide invested between $20 billion and $30 billion in artificial intelligence. This includes both internal R&D investment and stable acquisition flows. So far, technology giants such as Alibaba, Amazon, Baidu, Facebook and Google have invested more than three-quarters of their total investment in artificial intelligence. From 2011 to February 2017, these companies completed 29 of the 55 large M&A transactions in the US and 9 of the 10 major transactions in China.

Active investments are helping these companies gain access to key talent, technology and data sets, potentially creating barriers for their chronic competitors. Early adopters of artificial intelligence were able to use artificial intelligence as a unique advantage to enter the relevant industries. Artificial intelligence assistants like Siri, Alexa, or Cortana can serve applications in new areas such as healthcare and implement functions such as recommending hospitals or experts based on a person's unique medical history, or monitoring chronic disease indicators in real time. The realization of these possibilities will make the long-standing industry boundaries increasingly blurred.

Native digital companies also have their unique advantages. They understand the value of the vast amount of clean data that they get from their core business. They usually use agile "test and learn" methods in their operations. They have a clear understanding of how artificial intelligence technology can enhance its core business, whether it's Amazon's Kiva robot or Facebook's personalized robot. Existing traditional industry companies can adopt their technology, but will find it difficult to catch up with them.

In ASEAN, industry players in the high-tech industry (banking and telecommunications) who are most proficient in digital technology have begun to take action, but they are shorting. Many of ASEAN's telecom operators have begun to enter the field of artificial intelligence to optimize customer intelligence, but they have encountered difficulties in expanding their business, often because of their lack of key skills in data science and business translation. In the absence of bold interventions, participants in other industries are likely to encounter the same obstacles, forcing them to rely on professional technology providers. Perhaps the most striking example is IBM's Watson system (see Item 1).

Source 1: IBM Watson, the world's most famous artificial intelligence system

IBM has introduced the concept of artificial intelligence to the public through its "Watson" supercomputer. "Watson" defeated all human players in the American quiz show "Dangerous Edge". Since then, IBM has been demonstrating "Watson" cloud-based predictive analytics capabilities to customers in a variety of industries. In order to have the ability of humans to answer questions, Watson uses 80 trillion floating-point operations per second to access 90 servers with more than 200 million pages of data. It can mine text, perform complex analysis of massive unstructured data, and run the world's most powerful search engine.

Watson's basic cognitive computing technology is suitable for a wide variety of applications, including the following:

(1) Healthcare and medical research: Watson can process large amounts of patient data, look for treatment options that drug researchers may not expect, and then propose new hypotheses for further evaluation. Its processing power is being used to test patients and clinical trials, diagnose cancer and determine treatment options, manage chronic diseases, and accelerate drug development.

(2) Education: Artificial intelligence has great potential for personalized teaching to adapt to each student's learning style, and to ensure that students move to more advanced topics while mastering the current content. Watson can provide critical insights based on the demographic characteristics, strengths, and weaknesses of the students, enabling teachers to develop targeted, dynamic instructional programs.

(3) Public safety: “Watson” is deployed in the intelligent city control center to predict criminal activities and assist the public security department to allocate limited resources reasonably. It can now also assess resistance to cyber threats and take corrective action accordingly.

(4) Analysis of sports events: Analysis plays an important role in professional sports. Participants can analyze a large number of performance indicators and variables to gain competitive advantage. "Watson" was used to analyze a basketball team's game, determine the skill gap of the players, and recommend who should be signed, who should be in a specific situation

(5) Media broadcast: “Watson” has been able to automatically edit video highlights, the most recent one being in Wimbledon, which usually requires a complete content operations team to complete. With this technology, the turning points in the game can be captured immediately and automatically posted on different social media channels, resulting in a greater sensation.

Worldwide, the popularity of artificial intelligence is often related to the degree of digitization. In ASEAN, the pace of digital development is accelerating. In 2011, only 6% of Asian companies mentioned the terms “big data”, “advanced analysis”, “artificial intelligence”, “machine learning” and “Internet of Things” in their annual reports. By 2016, this ratio has reached one-third, indicating that these technologies are gaining momentum and are gradually becoming strategic priorities.

We found that in all industries, early adopters of artificial intelligence achieved higher profit margins than their peers (see Exhibit 4), especially in the manufacturing, financial services, transportation, and logistics industries. In order to consolidate the market and eliminate competition, most of these companies have given this residual value to customers. This "winners" situation has further exacerbated the "digitise or die" situation in which many incumbents are located.

Exhibit 4: ASEAN's artificial intelligence applications create superior profitability and a huge value base

However, the adoption of artificial intelligence has not achieved its maximum value. Pre-experiment and subsequent implementation require the company to make forward-looking and extensive observations on how artificial intelligence can be applied to its core business. For companies in the traditional non-technical industry, late implementation may be prohibitive. So far, high-tech, telecommunications and financial services companies have dominated the ASEAN countries. We have also seen a surge in public service activities such as transportation and health care, driven by a number of government agencies and the Smart City program in the region.

At the national level, Singapore is a leader in artificial intelligence experiments in a variety of industries. But countries throughout the ASEAN region have some initiative (see Exhibit 5).

Figure 5: The popularity of artificial intelligence in various industrial sectors within ASEAN countries

While these phenomena are encouraging, ASEAN will need a clearer business case and a stronger data ecosystem. In addition, if artificial intelligence wants to realize its full potential throughout the region, it requires more harmonized talent and skills.

Development process in various areas of ASEAN

Below we will study the application of artificial intelligence in some specific industries. We start with two industries that account for about half of all ASEAN current use cases for artificial intelligence: financial, high-tech and telecommunications. After that, we focus on the manufacturing and transportation industries, which have broad development value and two priority public service areas: health care and education, all of which are likely to bring significant benefits to society.

Financial Services

So far, financial services companies in Southeast Asia have mainly improved their customer experience through artificial intelligence. For example, Hong Leong Bank of Malaysia (Malaysia) uses the IBM® Watson system to understand customer sentiment by detecting how customers speak on the phone. DBS, based in Singapore, has opened a digital bank that uses virtual assistants to predict and answer customer questions. The Hong Kong startup CompareAsiaGroup, which operates in five ASEAN countries, uses machine learning technology to connect customers with financial, communications and utility services in Asia.

For artificial intelligence, to have a broad, long-term impact on an industry, these banks in Southeast Asia may need to refer to examples of artificial intelligence that have been successfully applied in the US and China. The application of artificial intelligence to functions such as credit scoring, dynamic pricing, and digital marketing has already demonstrated its value in many places, but few companies have expanded the scale of such applications in ASEAN. To seize this opportunity, banks need to constantly develop new skills, and financial technology start-ups must continue to innovate. Of course, first of all, these companies must accelerate their basic digital pace.

Digitizing customer interactions and establishing data collection, management, and analysis processes are all priorities, because artificial intelligence tools require large amounts of data. The business case that has completed this digital transformation further reinforces the fact that ASEAN's middle-class consumers are at the heart of the consumer base, they can use digital technology, and they often shop online and choose their own satisfaction. commodity.

Almost 300 financial technology startups have invested in this area to provide solutions for online payments, p2p loans and wealth management. In theory, embedding artificial intelligence technology into their products can make them powerfully occupy the city's excellent technical strength and the ability to design practical applications for artificial intelligence, creating value for customers and making their experience Smoother, this will make the best companies stand out. The development of this field will have a huge social impact. Approximately 266 million people in the ASEAN region lack credit channels. Ultimately, artificial intelligence will provide affordable financial services to vulnerable and low-income people who are often excluded from traditional banking systems. For example, in China, Alibaba has developed the company's internal Sesame Credit Service (Zhima?Credit) using advanced analytical tools and rich business and consumer data, which may open Alibaba's way of providing loans to small loan groups. door.

Government regulators can determine the pace of innovation for financial technology companies. Over time, they may also open up the banking platform to ensure that companies compete fairly on data access. It is crucial to have carefully weighed rules between data availability and privacy, just like India's Aadhaar (Biometric Identification System). Officials may choose to allow artificial intelligence to test data in a controlled environment.

High-tech and telecommunications industry

It is not surprising that high-tech and telecommunications companies are early adopters of artificial intelligence technology. Globally, some technology giants have developed artificial intelligence applications that disrupt the traditional physical industries, such as retailers (Amazon) and entertainment (Netflix).

Its excellent results are also surprising. Amazon has saved $22 million a year after acquiring a robotics company, saving operating costs by 20%. At the same time, Netflix estimates that its artificial intelligence recommendation tool helped it avoid the $1 billion annual unsubscribe service.

Similarly, many small-scale companies in Southeast Asia are constantly working hard. Local telecommunications companies are already at the forefront because they can use their extensive population coverage and access to data, and by 2020, 90% of adults in emerging countries will use mobile subscription services.

Telecommunications companies have long used analytics tools to predict customer churn and some long-term cross-selling of additional services. But now the possibilities are much bigger – including the opportunity to enter a new type of market. People who don't have a bank account today can get basic financial services through mobile devices, and the data generated by their transactions can be used by potential customers of banks to identify insurance and other financial services such as loans. Telecom companies are also using artificial intelligence to enter other industries. . Singtel has established a data analysis subsidiary to collect, model and visualize shopper data, while India Telecom's analytics division focuses on retailers' digital marketing and bank credit scoring. ASEAN has also promoted the rise of small high-tech startups that are supported by a growing venture capital ecosystem.

By definition, the high-tech industry has intersected with all other areas involved in artificial intelligence – so terms such as financial technology, medical technology, and educational technology have also been widely disseminated. The government supports local innovators to gain certain benefits, and these innovators can pave the way for a wider spread of artificial intelligence. By improving computer science education, the government can narrow the key gaps in professional high-tech job skills, develop regulations, promote the use of anonymous data, and encourage the collection of cross-industry and cross-domain data. Singapore has taken some measures in this regard, such as encouraging entrepreneurship and providing a large amount of government grants. This move has enabled Singapore startups to succeed in the 2017 Global Entrepreneurship Ecosystem Report, ahead of Osmarks in Texas and Stockholm. In addition to establishing a startup, the government's support can also add a certain degree of visibility and prestige to domestic companies, thus retaining those who might have flowed overseas. This is also a topic we should pay attention to when discussing artificial intelligence.

manufacturing

Artificial intelligence technology will play an important role in the next phase of development in the industry. Companies will soon be able to manage the plant floor in real time and connect the entire value chain with seamless data streams for real-time decision making and increased productivity. This new world of digital manufacturing is often referred to as Industry 4.0.

In ASEAN countries, the use of artificial intelligence and the Internet of Things is a natural process. The largest companies in the region may become leaders because their business scale has already involved the areas with the greatest potential benefits. Thai food and beverage group ThaiBev and Malaysian car manufacturer Proton are just two of the major brands that aim to bring Industry 4.0 technology to their factories.

Source 2: ASEAN's Artificial Intelligence Technology Startup

In 2016, the region's total venture capital reached $2.6 billion, an increase of about 60% over the previous year. In addition, the lag in economic development and increasing social problems provide opportunities for the development of technology-driven solutions.

Many tech entrepreneurs are developing artificial intelligence technologies and applying them to local instances. These regional startups do not have the resources or talent pool that international technology giants like, but they also illustrate the importance of finding market opportunities and designing local business models in the local area.

Examples of artificial intelligence based technologies used by ASEAN startups are as follows:

Natural language processing

(1) Myanmar's Bindez uses natural language processing and text analysis to track hate speech on the Internet.

(2) Indonesia's Kata.ai is developing a Malay language processing algorithm, Malay is the main language of more than 250 million people in Indonesia and Malaysia.

(3) In Vietnam, FPT designed an artificial intelligence platform to help application developers automatically interact with end users based on natural language processing interfaces. Potential applications for such platforms include call center chat bots, virtual agents, and related voice applications.

Machine learning

(1) Network security startup Cloudsek is committed to providing machine learning-based solutions that help companies identify and address cyber threats in real time.

(2) In Indonesia, Ruangguru is exploring ways to implement personalized education services through machine learning, using the vast amount of academic data it has.

Image Identification

(1) Sero, a Vietnamese agricultural startup, provides crop intelligence to farmers through artificial intelligence analysis of images and field data.

Source 3: What is Industry 4.0?

“Industry 4.0” is a term used to describe the digital transformation of manufacturing, with the aim of combining a range of new technologies with manufacturing. The Internet of Things, artificial intelligence, robotics, and 3D printing technology can turn a factory floor into a flexible, self-maintaining operating system. Sensors can stream continuous real-time data streams into machine learning algorithms that remotely adjust complex systems, processes, and machines. These same types of systems can be used to coordinate the entire supply chain and monitor customer usage to inform future product designs and new services.

According to several studies, McKinsey estimates that Industry 4.0 can increase production efficiency by 15% to 20% in manufacturing. Excellent global manufacturers in Germany and other places have successfully demonstrated their feasibility and commercial value.

(1) Predictive maintenance: Applying machine learning tools to data collected by IoT sensors enables manufacturers to predict equipment failures and prevent machine damage and downtime through preventive maintenance. Some companies have managed to reduce overall maintenance costs by 10%.

(2) Increase in output: Industry 4.0 technology enables manufacturers to optimize the use of raw materials and increase production. An artificial intelligence semiconductor manufacturing system reduces the scrap rate of scrap metal by 30% by connecting thousands of variables to the machine group and sub-processes.

(3) Product design and after-sales service: Smart products, such as smart cars, can feed customer experience data to the product manager. This capability opens up new ways of service and is reflected in improved product design.

But many manufacturers are still hesitant considering the up-front capital investment required for modern plants and the cost of digitizing large amounts of tangible assets. Because ASEAN's labor costs are low, companies do not always see business reasons for changing their business practices.

However, in the long run, this cost calculation may change. As the region develops and the population ages, labor costs may rise, thereby reducing the size of the available workforce. China's manufacturing wages have doubled in the past decade, and Chinese companies have begun to actively adopt robotics; in fact, they are expected to invest $59 billion in machine automation by 2020.

Policymakers in the region can encourage digital transformation of manufacturing as a top priority for productivity growth, thereby promoting economic growth across the region. For example, the Singapore government supports the launch of McKinsey's Digital Competence Center (DCC), which focuses on Industry 4.0, in Singapore. Singapore DCC has established a partnership with the advanced Advanced?Remanufacturing?and?Technology?Centre (ARTC) to introduce new technologies to manufacturing companies and help them develop new capabilities. As part of the broader economic transformation blueprint, Malaysia and Thailand also include Industry 4.0.

Transportation and logistics

Rapid urbanization is putting pressure on transportation systems in cities around the world. And to solve this problem is costly: in Asia alone, the direct cost of traffic congestion is about 2% to 5% of GDP. Most of the world's major cities are struggling to solve problems related to rapid urbanization. They plan a smart city blueprint that integrates artificial intelligence and the Internet of Things to improve network efficiency by managing the infrastructure in a “smart” way.

By 2030, most cities will adopt new automotive technologies such as car sharing, autonomous driving and electrification, although these technologies will not succeed at the same time. In the future, in the most densely populated cities, “seamless movement” may be able to be implemented in the crowd, and can be brought home from home to destination. “Seamless Movement” will rely on a combination of autonomous driving and shared vehicles, complementing smart, integrated public transport infrastructure (smart cars and buses, subways and traffic management).

Private companies can play a role in realizing this vision of seamless mobility.传统汽车制造商和谷歌、百度等高科技巨头正斥资数百万美元投资自动驾驶汽车，采用防撞和路线选择优化系统，以提高安全性和降低燃料消耗。福特已经从一家汽车制造商转型为一家“机动车”供应商。该公司已经成立了一个城市解决方案部门，该部门将利用人工智能技术无缝整合许多移动设备，从公共交通到出租车再到共享单车。

新加坡是东盟在执行其“智能移动2030”计划时的领先者，该计划要求人工智能系统做到实时管理列车、公交车、汽车和自行车交通。马来西亚的雪兰莪州也在推行类似的计划，以及印度尼西亚、菲律宾和柬埔寨的智能城市项目也正在进行中。

初创科技公司正在成为这一领域的重要组成部分。Yogee网销售使用了机器学习技术的灵活管理软件，因此它变得更加智能，使用的范围更广。在7个东盟国家运营的叫车平台Grab，已聘用了200名工程师和数据科学家，专注于利用人工智能改善客户服务，并进一步优化其司机队伍。

城市政府面临的紧迫挑战是与战略行业参与者和科技创业公司建立合作关系。然而，这些合作的整合是相对复杂的。当然，城市的净效益是显而易见的，比如减少拥堵和提高了安全性。但要调整私人投资和公共奖励的激励机制是很有挑战性的。此外，大多数东盟国家都专注于自动收费站，而且对大型公共投资兴趣不大。尽管面临诸多挑战，但在过度拥挤的东南亚城市中改善生活的主要潜力，使得建立高效的公私伙伴关系变得至关重要(我们将在最后一章回到这个话题)。

医疗保健

在全球范围内，人工智能已经以多种方式不断展示出改善医疗服务的潜力。深度学习可以让机器查阅大量有关疾并治疗和结果的数据，从而快速找到可以改善诊断方案和病人护理的见解。IBM利用其人工智能支持的Watson超级计算机，让医生可以在几秒钟内筛选数百万页的医学证据，从而为患者设计出最优的癌症治疗方案。可穿戴机器人设备可以远程追踪病人的健康状况，并且带有提醒功能，可以叮嘱病人及时吃药。虚拟代理已经在分析放射学和肿瘤报告，并为病人提供建议。

MGI之前的一项研究估计，在医疗保健领域扩大数据的使用每年可以产生超过3000亿美元的价值，其中三分之二来自于将国家医疗支出减少的8%。

医疗保险是另一个有潜力的储蓄领域。从全球来看，机械制造解决方案优化了索赔处理、减少了欺诈和改善了健康状况预测，这可能会带来更好的预防保健和更低的索赔。

在东盟，在病人护理领域广泛采用人工智能的做法还需要数年时间，但现在已经出现了几个成功的例子。新加坡政府机构IHiS（集成健康信息系统）旨在创建一个全国性的企业分析平台，汇集和分析来自多个医疗保健系统的患者数据，并生成有助于改善治疗结果的见解。通过提供在线医生咨询和可穿戴式传感器引导的家庭诊断，这可能会使管理慢性病变得可行。其次的好处包括尽量减少事故和急诊单位的过度拥挤，以及减少病人的医疗费用。像Holmusk这样的初创公司也在为特定的病例开发数据和应用程序，比如糖尿玻在越南，ViCare保健应用程序在Facebook?Messenger上为病人提供了一个聊天机器人，可以为病人回答一些基本问题。

拥有大量人口但没有足够多的医生和专家的国家将从这些技术中获益最多。IBM的“沃森”也许可以在印尼提供服务。2014年，印尼只有41名放射肿瘤学专家，却要为2.5亿人提供治疗，而且这个国家这一年因癌症死亡近20万人。

然而，该地区没有足够的整合数据来支持先进的分析技术，更不用说人工智能了。医院有数据，但通常是以纸质形式来记录的，想要共享比较困难。大多数东盟国家要求数据不可以留出国外，这就限制了建立区域性数据库的机会。更重要的是，将病人数据集中在一起，并将其开放给机器学习，即使是以匿名的形式，或者将使用可穿戴设备的要求捆绑到保险折扣上，也可能与隐私规范和法律不一致。

医院和保险公司将决定药品如何使用人工智能。但是，与传统银行一样，医院和保险公司在转变组织的过程中也面临着挑战，不仅是通过积累数据，还要通过提高他们的数字化能力，将技术整合到他们的工作流程中，以改变他们的文化。创新可能来自数字化本土公司。医疗保健公司可以通过赞助有前途的创业公司来与这些公司结盟。新加坡的一些公司已经采取了这种做法。政府可以通过提供有关数据共享的监管指导，以及在需要的时候提供公共投资，从而促进这一过程。

教育科技已经是一个蓬勃发展的领域，为人工智能扎根提供了肥沃的土壤。与金融科技一样，教育科技也迎合了一个巨大的市场：全球教育支出占全球GDP的近5%。投资者注意到，一家投资银行预测，到2020年，教育科技投资将增长至2500亿美元。

人工智能在课堂上的潜力让人兴奋不已。例如，以人工智能为基础的智能家庭教师系统（ITS）旨在提供大规模的一对一教学。这些聪明的导师可以追踪每个学生的表现，找出学生觉得困难的概念，并为每个人找出适合自己的学习方法。人工智能还可以减轻教师的一些日常工作，给他们更多的时间来教学。一位乔治亚理工大学的教授在一个学期内使用了一个人工智能教学助理，处理来自他在线课程的1万多个问题。人工智能助手还可以从事更智能的工作，如评分和记录分数，使教师能够专注于更有创造性和更具附加值的工作。

其中一些技术已经在东盟地区得到采用。新加坡和马来西亚的大学已经试验了预测软件，以指导能够防止辍学的干预措施。但是，东盟还有很长的一段路要走，才能对其产生重大影响。大多数成员国都没有收集能让人工智能算法得出结论并做出预测的综合数据。该地区的许多地区也缺乏关键的IT基础设施。2016年，只有不到一半的亚洲人口使用互联网，其中包括大多数东盟国家的多数人口。

东盟国家可以首先利用现有技术，更易于实施的方法，以改善教育的质量和公平性。像可汗学院(Khan?Academy)或马来西亚亚洲电子大学(Asia?e?university)这样的在线自学课程提高了入学的机会。通过配备预装材料和低带宽通道的设备，在偏远地区或缺乏熟练教师的地方，教育质量和公平性得到了改善。

这些工具并不能保证更好的教育成果。政策制定者和地方行政官员必须调整政策，以满足学生的实际需求，并切实地考虑基础设施的准备和规划。教育科技解决方案应该专注于教学，将技术解决方案与现场教学的优势结合起来，并与本地适用的课程相匹配。建立一项能够评估国家系统可行性和性能的教育科技政策，将允许各国在时机成熟的时候充分利用人工智能。

类似地，各国现在可以开始为人工智能技术的发展做准备，开发更完备的国家数据库，更先进的技术解决方案依赖于此。这包括获取学生人口统计数据、环境变量、出勤率、学校属性、个人、学校和地区关系的数据。政府不需要自己收集和整理数据；他们可以与国际或当地公司合作。然而，政府需要参与其中，因为它们往往是主要的数据收集者，必须确保数据隐私。

一旦这些数据结构就位，机器学习算法——包括那些在该地区以外开发的算法——就可以在国家层面上学习。这将为教育部门提供如何部署教育资源和调整政策以满足劳动力需求的宝贵指导，目前还没有哪个东盟国家能够实施。在个人层面，国家层面的数据可以支持并指导教师、家长和管理者如何让学生留在学校，以及采取什么样的干预措施来降低学生失学的风险。

解决跨领域的挑战和机遇

正如上面讨论的行业例子所示，人工智能可以极大地提高生产力。如今，企业可以使用强大而成熟的分析工具，从而提高运营绩效，创造新的市场机遇。

但这并不是一个简单的命题——没有一个单独的组织能够独自解决围绕这些技术的所有问题。有复杂的伦理、法律和安全问题有待回答，而最终对就业的影响仍有待观察。整个东盟地区将需要加强其数字基础设施建设，发展拥有先进数字技能的更大的人才库，并确保建立一个经过深思熟虑的监管框架。正如我们在下面第3节所讨论的，解决这些问题需要公共和私营部门的合作和共同努力。

今天，东盟的大部分地区在数字普及方面落后于其他国家。但这并不是该地区的公司认为下一代技术与本土市场没有相关性。事实上，一些技术欠发达的地区可能孕育着一些最有前途的机遇。它们可以从一个全新的领域开始发展，它们不太会被遗留系统和规章制度所困。灵感可以从中国获得，中国在非常短的时间内成功建立了一个强大的数字生态系统——而在欠发达经济体中，东盟的初创企业也可能会蓬勃发展。

正如其上所述，东南亚不同业务领域的数字化成熟程度各有不同。如果单纯依靠市场的推动力量，金融服务业、高科技和电信行业的先驱者们或将最先接纳人工智能。然而，要抓住人工智能的市场价值，并真正改善社会并非易事。这将需要政策制定者的结构性干预措施，加之行业参与者的积极承诺和践行。

以下我们将列举一些该地区在人工智能发展中需要解决的关键问题，同时探讨政府和企业可以在其中发挥的重要作用。

对于所有人工智能的发展潜力来说，在没有人类指导的情况下让机器进行学习和做出决策，并对其进行管理是一项艰巨的任务，也是一种重要责任。这些技术正在把整个社会带入未知的发展方向。尽管我们知道，人工智能应用程序的增长需要基于数据生态系统和数字能力的某些基本要素，但我们不知道人工智能技术进过第二和第三次迭代后会出现何种商业案例，也不知道公众态度会发生何种转变。人工智能的普及还涉及到一些社会价值观的问题，但这些问题没有任何确定性的答案。因此，我们提出一些开放性的问题，从而引入更多的深入讨论。

1、私营部门的发展路线

对于企业来说，人工智能的普及遵循了其他数字技术发展的路线图。这些元素包括明确定义的用例或价值源;健壮的数据生态系统;熟练使用系统和工具的雇员;与核心业务的工作流进行有序整合;以及接受“测试和学习”方法的开放文化。对于整个东南亚的企业来说，即使是在前沿行业，其中的数据生态系统、运营文化和关键技能往往都存在着不少障碍。

创建健壮的数据生态系统

对于人工智能技术来说，必须有稳定的可靠、可操作和安全数据，这是人工智能技术学习和完善功能的基本方式。但东南亚地区的多个行业在数据基础方面存在很多困难。目前其中的许多行业都缺乏足够的关键传感器系统来跟踪操作数据。在某些情况下，人工智能程序需要依托实时数据流进行决策和相关操作。例如，东南亚的多家电信运营商将实时网络数据传输到他们的数据库中，并利用这些数据来开展与客户密切相关的活动和通知。举一个简单的例子，当用户在接近他的数据流量上限时会收到相应通知。但只有少数几个行业将这种类型的解决方案实现规模化应用。

即使很多公司设置了足够的传感器，但其中很多依旧缺乏合适的基础设施来存储数据，更无法将其聚合成可操作的数据形式用于相应决策。在许多公司中，数据存储都是各不相关的孤岛。在另一些公司中，人们收集了大量的数据，但从未进行有效分析。麦肯锡的一项研究发现，石油钻井平台上3万个传感器捕捉的全部数据中，只有不到1%被有效利用。

现在，随着基于云数据管理平台的出现，存储和分析数据的成本正在不断下降，数据使用的便利性也在不断提高。许多中小企业和创业公司都采用这些新技术平台来降低成本（见资料4：“整合数据策略需要建立一个强大的数据生态系统”）。为了实现他们关于人工智能的目标，参与者需要积极拥抱这些新技术，同时确保正确的数据管理能够在业务便捷性和规模化之间实现平衡。

管理风格向数据驱动过渡

在企业中实施人工智能所需要的最根本文化和组织转变，就是要接纳数据驱动决策。曾经凭直觉做出的决策现在可以基于证据而做出，甚至可以是自动化的。由于人工智能在东南亚地区仍是一个相对较新的概念，企业也需要逐步适应这种新的模式。

即使是那些对数据手机和数据分析进行投资的公司，也可能无法在决策过程中有效使用数据。其中包括以下一些问题：

（1）对业务情况和价值来源的表述不到位，导致决策基础薄弱。

（2）中层管理人员缺乏相应的能力建设，不愿意依靠人工智能的分析作为决策的依据。

（3）对雇员特别是对一线工作者的再培训投资有限。

（4）缺乏雇员引入机制。

（5）与所有文化转型一样，领导力对于人工智能的成功实施至关重要。麦肯锡全球研究所的一项调查发现，那些成功部署了人工智能技术的公司受访者表示，相比于那些没有采用任何人工智能技术的公司，其高管层的支持度几乎是其他公司的两倍。

资料4：整合数据策略需要建立一个强大的数据生态系统

数据正在成为一种新的资本形式。跨行业研究显示，平均而言，在决策过程中仅有不到一半的组织结构数据被用于决策，超过70%的员工获得的数据是不必要的，而数据分析师80%的时间是用于发现和准备数据的。

企业需要采取一种程序化的方法来构建数据资产，并在所有业务部门的支持下，利用这些资产来改变整个企业。以下是这类数据驱动转变的三个关键组成部分：

一个清晰的数据战略，往往与企业的愿景紧密相连

（1）第一步是弄清数据如何被用来推动关键的业务目标和文档用例的实现。

（2）下一个问题是，要确定企业数据的关键缺口，需要用新的集合系统或互补的外部数据加以填补;企业也应该对提供独特优势的专有数据资产保持开放的态度。

（3）将简单的成本效益分析与每个用例联系起来，有助于评估它们对业务的重要性，并指导诸如“外采或开发”之类的决策。

数据架构和路线图实现的总体蓝图

（1）数据架构的设计源于符合公司需求的数据模型视图以及优先级用例。

（2）该架构的设计目标是优化数据收集、聚合、使用和后续更新，同时保持数据准确性和一致性，确保数据的安全性。

（3）选择合适的技术能够控制升级系统的成本，同时为系统运行提供足够的灵活性。

达到持续性决策和丰富数据集的有效数据治理

（1）数据治理机制的本质是选择集中的、联合的、或完全去中心化的数据组织，以及首席数据官在核心管理中的位置。

（2）根据数据及其来源的重要性，定义与外部各方的数据进行交互和共享的规则。

（3）制定了相关的指导方针，以开发能够对数据进行阐释的硬资产，比如企业数据词典和监控数据质量的仪表类应用。

打造正确的技能组合

各个公司均表示，在试图将数据和相关分析整合到现有业务的过程中，找到合适的人选是他们面临的最大障碍。

麦肯锡最近的一项调查显示，大约有一半的企业高管认为招募一名合格的数据分析人才难上加难。尤其是对数据科学家的需求量更大。而恰恰这些人就是设计、开发、部署和培训人工智能技术的人。目前这类人才非常短缺，即便是在像硅谷这样的全球性人工智能中心也是如此。而东南亚这类人才的短缺更为严重。

另一个同样重要的角色是商业翻译，他们可以充当分析人才和实际应用之间的纽带和桥梁。除了精通数据，商业翻译还需要具备深厚的组织架构知识、行业方面或业务方面的专长。他们能够向数据科学团队提出正确的问题，并从他们的分析中获得正确的见解。

当然，公司也可以选择把数据分析业务进行外包，但对于商业翻译这种角色来说，其可以利用自己的专有知识深入组织架构的内部。而很多企业所需要是从内部打造相应能力。对于企业来说，其中一种选择是“构建-操作-转让”模式，即来自外部专业公司的专家被整合进跨职能项目团队中。这些专家会与内部员工进行紧密合作，其向员工提供关于如何与人工智能技术系统进行合作的诀窍，同时员工会利用自身的运营经验来加深专家对公司真实需求的理解。而员工也相应获得了新技能，能够在初始阶段之后不断自我提高和完善。

2、政策制定者面临的结构性挑战

目前，东南亚的政策制定者需要通过合理政策将现有创新转化为可持续增长。政府可以通过建立坚实的政策基储设定有抱负性的目标、刺激私营部门的创新并采纳人工智能来推动这一转化。

支持开发和采纳人工智能的政策

东南亚可以通过地区政策而非本地化政策来推动人工智能的发展和普及。最重要的任务之一是建立一个开放但安全的数据环境，这是数字以及人工智能技术的生命线。我们的研究显示，东南亚地区的流动性具有高度的全球联系，包括商品和服务贸易、人员流动和资本流动都是如此。但在跨境数据流动方面，东盟的全球联系明显较少（表6）。构建该地区的数字基础设施是关键的一步，而数据治理是其中的核心组成。

跨太平洋伙伴关系协定（TPP）为东盟解决数据交流障碍提供了一个机会，而且它提出的一些框架可以在地区层面进行考虑。 These include:

（1）制定标准，保护消费者不受网络诈骗的侵害，并明确个人信息将如何跨界交流。

（2）防止和应对不断变化的网络安全威胁。

（3）保护数字知识产权，同时减少海关、互联网接口、产品歧视等对在线商务造成的障碍。

（4）避免“数据保护主义”，规范企业数据存储。

政府也可以通过让自己的公共数据更易于访问，从而建立更加开放的数据生态系统。这可以为第三方应用、人工智能开发者和创业公司提供丰富的开发模块。

随着人工智能不断出现新用途，政府和整个社会也需要努力规范数据隐私的原则。如果政府和企业收集的数据被匿名化，公众还有权知道他们的数据是如何被使用的吗？那些人工智能的用户有义务去解释他们的机器是否符合公众利益或个人幸福吗？

各国政府还必须考虑自身在解决技术颠覆带来的负面效应方面所发挥的作用。其中一项主要战略将是长期投资教育，其中也包括继续教育体系，从而帮助处于职业生涯中期的劳动者跟上数字经济不断变化的需求。但这同样会引发许多问题。政府如何确保妇女和农村人士能够平等地接受数字化培训？他们能否在一定程度上抵消数字颠覆带来的不平等扩大等风险？哪些行业最适合被颠覆？政府和公司应该如何分配再培训的责任？人工智能技术本身能够提供部分解决方案吗？

图表6：东南亚地区的联系、数据流量以及人均国内生产总值的排名。

在东南亚国家之间，基于贸易和资金流的传统领域高度互联，但相关之间的数据流联系却不那么紧密。

政府可以利用财政政策来解决失业和社会混乱问题。但除了提供资金的安全保障外，还有其他方法可以利用技术来限制失业吗？如果一个由人工智能推动的经济需要更少的劳动力，那么是否有可能通过设计让工作安排更加灵活，让公司能够协同工作？

最后，由于早期采用者紧握人才、兼并更小的创新者、并获得不成比例的经济利润，人工智能行业存在着市场垄断的风险。但这种可能性目前被大型跨国公司在该地区的技术扩张以及普及所带来的益处所抵消。

当政府通过监管或财政政策进行干预时，应

Windows Tablet

The latest Windows has multiple versions, including Basic, Home, and Ultimate. Windows has developed from a simple GUI to a typical operating system with its own file format and drivers, and has actually become the most user-friendly operating system. Windows has added the Multiple Desktops feature. This function allows users to use multiple desktop environments under the same operating system, that is, users can switch between different desktop environments according to their needs. It can be said that on the tablet platform, the Windows operating system has a good foundation.

Windows Tablet,New Windows Tablet,Tablet Windows

Jingjiang Gisen Technology Co.,Ltd , https://www.gisentech.com