Literary Insights

“If you liked the book, you can purchase it using the links in the description below. By buying through these links, you contribute to the blog without paying any extra, as we receive a small commission. This helps us bring more quality content to you!”

BOOK LINK:

CLICK HERE

Artificial intelligence (AI) has recently become a topic of great interest and importance globally. Even young children are asking questions about how AI will impact the future.
China has experienced a surge of excitement and investment around AI over the past 3 years. Chinese AI companies and researchers have made major advances, turning China into an AI superpower alongside the United States.
How these two countries choose to compete and cooperate on AI will have major implications economically and politically.
Predicting the future of AI is difficult because it depends on human choices and actions. The author aims to shed light on how we got here and inspire conversations about where we go next.
This story is about both intelligent machines and human beings. Our AI future will reflect the values and wisdom we bring to guiding it.
In 2016, Google’s AlphaGo AI defeated top Go player Ke Jie in a highly publicized match. This was a “Sputnik moment” for China, lighting a fire under the Chinese tech community to pursue AI with immense resources and determination.
AlphaGo uses deep learning, a breakthrough AI technique that is enabling machines to match or surpass human capabilities in many domains. This is bringing the AI revolution long promised.
Deep learning poses a threat of widespread job displacement across industries and skill levels. The author saw this as a more pressing concern than Skynet-style killer robots.
But the author also saw hope in Ke Jie’s emotional reaction to defeat. It showed the enduring power of human qualities like competitive spirit, love of an art form, and connection between people. Finding meaning through such connections can help humans adapt to an AI future.
China is poised to catch up to or even surpass the US in AI given its massive investment, research, and entrepreneurship in the field. But more importantly, AI can be an opportunity for people worldwide to rediscover our shared humanity.
I was born in Taiwan but moved to the US at age 11, finishing school there. After college at Columbia, I pursued a PhD in AI at Carnegie Mellon in the 1980s, drawn by the grand goal of understanding intelligence.
At the time, AI research was divided into two camps - rule-based (“symbolic”) systems vs neural networks. Rule-based systems encoded human expertise into logical rules, while neural networks mimicked the architecture of the brain.
Neural networks showed early promise but were eclipsed by rule-based systems in the 1960s-70s. They had periodic resurgences but suffered from lack of data and computing power.
The recent AI renaissance has been driven by the success of deep learning, a breakthrough in training multi-layer neural networks. This was enabled by growth in data and computing power, and key innovations in efficiently training neural nets.
Deep learning excels at narrow AI tasks like computer vision and speech recognition. Its core strength is pattern recognition - finding correlations in massive data sets to optimize outcomes.
Tech giants like Google raced to hire the few deep learning experts and apply the technology. China was largely absent from the origins of deep learning due to isolation from global AI research.
Progress in AI has largely taken place in the United States, Canada, and the UK. Recently, Chinese entrepreneurs and venture capitalists have begun investing more in AI, but China was late to the deep learning revolution.
Many believe the US has a commanding lead in AI research that will only grow due to its talent, culture, and tech companies. However, this view overlooks changes in China’s tech environment and misunderstands the AI revolution.
We are moving from an age of discovery to an age of implementation, where deep learning is applied to revolutionize industries. This benefits entrepreneurs more than elite researchers.
We’re also moving from an age of expertise to an age of data. With enough data, algorithms by average engineers can outperform those by top researchers.
These transitions play to China’s strengths (entrepreneurs, data) and downplay weaknesses (cutting-edge research). This gives China an advantage to lead in AI implementation.
China’s internet entrepreneurs are hardened by an ultra-competitive environment. This produces a relentless focus on execution and innovation that exceeds even Silicon Valley.
In sum, factors like data availability, entrepreneurial culture, and supportive policy environment favor China over the US in the AI age. The US sparked deep learning, but China will be its biggest beneficiary.
China is poised to overtake the U.S. in developing and deploying AI due to several factors: its entrepreneurs and abundance of data from its unique tech ecosystem. This could translate into huge economic gains for China.
The Chinese government is actively supporting AI research and development, unlike the U.S. government. This further tips the scales in China’s favor.
The rise of AI will lead to massive job losses and increasing inequality both within and between countries. By one estimate, AI could replace 40-50% of U.S. jobs in 15 years.
AI will concentrate wealth among a few AI tycoons as industries trend toward winner-take-all economics. This will exacerbate inequality.
A new bipolar world order is emerging with China and the U.S. far ahead of other countries in AI. This could create a divide between AI haves and have-nots.
The coming AI-induced economic and social crises dwarf the shifts in geopolitics between major powers like the U.S. and China.
Wang Xing became known in China’s early internet days for serially copying popular American internet companies like Friendster, Facebook, Twitter, and Groupon. This cloning was seen as shameless copying by Silicon Valley.
However, Wang’s success with Meituan, which started by copying Groupon’s model, shows that mere copying was not the key. Meituan succeeded by competing fiercely against thousands of other Chinese group buying sites to optimize and adapt the model for Chinese consumers.
The Chinese internet ecosystem had become a brutal arena where hundreds of copycats battled to the death. The competition forced innovation in product features, costs, execution, and developing business advantages. Mere copycats could not survive.
While the cloning may have seemed distasteful, it helped develop China’s digital skills. And the resulting intense competition produced world-class companies like Meituan through rapid innovation and iteration.
Analyses that dismiss China’s internet companies as just being protected copycats misunderstand the competitive dynamics at play in China. The battles in China’s “internet coliseum” forged strong innovators like Wang Xing.
China’s competitive startup environment forged tenacious entrepreneurs adept at quickly copying and adapting business models to make money. This will be a core asset as China builds an AI economy.
Silicon Valley startups are often mission-driven, focused on changing the world with innovative ideas. Chinese startups are market-driven, flexible and willing to enter any business that can make money.
These contrasting cultures stem from different histories - Silicon Valley’s abundance enables big thinking, while China’s scarcity mentality demands seizing any opportunity for profit.
China’s history of memorization and copying is culturally accepted, unlike in the Silicon Valley ethos.
China’s internet ecosystem produced copycat products that were mocked in the West. But it forged sharp entrepreneurs that will fan out across industries, applying AI to solve problems and make profits.
If AI is electricity, China’s entrepreneurs will rapidly electrify industries and deploy practical applications globally, especially in developing markets.
In the late 1990s, early Chinese internet companies like Sohu copied American models like Yahoo. This copying was seen as harmless flattery by Silicon Valley at the time.
However, Kai-Fu Lee, as head of Google China, experienced firsthand the threat posed by precision copying when a fake search engine copied Google’s look so closely it caused a public relations crisis.
Some viewed early Chinese copycats as lacking innovation, just copying the exterior form but not the substance of American tech firms.
However, Lee saw the copying not as a stumbling block, but as a necessary building block. China lacked the educational and commercial infrastructure to nurture innovation. Copying taught Chinese entrepreneurs valuable skills in design, architecture, and development.
The crude copying process gave Chinese companies experience in areas like user satisfaction and monetization. This copying phase laid the foundation required for true innovation to emerge later.
Chinese internet companies like Alibaba and Baidu beat Western competitors like eBay and Google in China through rapid iteration and localization.
They tailored products to match unique Chinese user behaviors and preferences, whereas Western firms stubbornly stuck to a rigid global model.
For example, Alibaba’s Jack Ma offered free listings to build trust and a user base, then monetized with premium services. eBay arrogantly insisted “free is not a business model.”
Baidu optimized search results for Chinese users who browsed more promiscuously like shopping, versus Western users who targeted specific information.
Western firms were slow to adapt and localize due to centralized decision-making and prioritizing global product elegance over localization. This allowed Chinese firms to drive a wedge between local users and Silicon Valley.
The successes of Alibaba and Baidu in China demonstrated how localization and rapid iteration could beat Western tech giants wedded to rigid global models ill-fitted for unique local user habits.
American tech giants like Google, Amazon, and Uber have failed to win over the Chinese market, while Chinese companies like Baidu, Alibaba, and Didi have thrived.
Western analysts attribute this to Chinese government protectionism, but the real reason is Silicon Valley’s flawed approach in China.
American firms don’t commit the resources needed to deeply localize and tailor products for China. They see China as just another market rather than needing reworked products.
Top local talent in China join domestic startups rather than American companies’ Chinese branches, where they hit career ceilings.
Chinese startups are intensely competitive, with entrepreneurs like Zhou Hongyi embracing a ‘gladiatorial’ ethos.
Tactics like copying competitors, smear campaigns, and even detaining rival executives are common in the Chinese tech arena.
The gladiatorial battles between Chinese startups forged a generation of entrepreneurs and turned the ecosystem ultra-competitive. This, more than protectionism, accounts for the rise of China’s tech giants.
Wang Xing founded a group buying site called Meituan in China in 2010, entering a crowded market with thousands of Groupon clones.
Having previously copied Facebook and Twitter, Wang had learned how to build tech products and compete fiercely.
While competitors burned cash on ads and subsidies, Wang focused on optimizing his product and operations to keep costs down.
Meituan avoided offline advertising and subsidizing users, instead iterating quickly based on data and optimizing the backend technology.
This “lean” approach allowed Meituan to survive as other group buying sites went bankrupt from excessive spending.
Meituan sharpened its operations and business model during years of intense competition to emerge dominant.
The group buying war in China forced entrepreneurs to rapidly iterate and execute at the highest level, incubating the “lean startup” methodology.
Wang Xing evolved from a copycat to a hardened gladiator through this hyper-competitive market.

Here are the key points from the passage:

Guo Hong is a Chinese government official who thinks like an entrepreneur. In 2010, as head of the Zhongguancun technology zone in Beijing, he wanted to turn it into a hub of Chinese innovation, rather than just a market for electronics and pirated software.
Guo visited Kai-Fu Lee at his newly founded incubator Sinovation Ventures. Lee had left Google China to nurture early-stage Chinese startups, sensing a new energy in the ecosystem as entrepreneurs applied their skills to local problems and the mobile internet.
Lee advised Guo that Zhongguancun needed easier registration processes and tax breaks to attract startups. It also needed Rolex buildings - cheap renovated factories providing space for young companies.
Guo implemented these suggestions, also establishing Zhongguancun as a special economic zone with subsidies and incubation programs. He essentially turned part of Beijing into a Silicon Valley for China.
The reforms paid off, as Zhongguancun became home to China’s biggest tech companies and attracted $12 billion in venture capital by 2015. Guo’s entrepreneurial thinking as a government official catalyzed the growth of China’s startup ecosystem.
Around 2013, the Chinese internet began morphing into an alternate universe distinct from the Western internet, with its own raw materials, systems, and laws.
Chinese tech companies like WeChat evolved beyond copying Western counterparts into indigenous innovation.
Key building blocks included mobile-first users, WeChat’s super app status, and mobile payments transforming phones into digital wallets.
This allowed pioneering of online-to-offline services, cashless environments, and intelligent bike-sharing.
Massive government support further turbocharged this innovation ecosystem.
Chinese companies embraced grunt work operations in the real world unlike aloof American platforms.
This is laying the groundwork for Chinese leadership in AI implementation, due to amassing huge quantities of high-quality real-world data.
The messy real-world focus gives Chinese AI more eyes into daily life compared to Silicon Valley’s online data.
China’s data advantage was an unintended windfall, not a master plan, but it positions the country well for the age of data-driven AI.
Google’s departure from China in 2010 created an opportunity for Chinese startups to build products for the emerging mobile internet space.
Sinovation Ventures invested in early mobile startups, some of which were acquired by Baidu, Alibaba, and Tencent (BAT). This helped BAT transition to mobile.
In 2011, Tencent launched WeChat, a messaging app that pioneered an “app within an app” model and grew incredibly quickly by adding voice, video, payments and other functionality.
In 2014, Tencent launched digital red envelopes on WeChat for Chinese New Year, getting millions of users to link bank accounts and use WeChat Pay. This was seen as an attack on Alibaba’s Alipay.
WeChat became a “super app” that dominated users’ digital and real lives, blurring online and offline worlds. This was a key piece of China’s alternate internet universe.
Chinese startups and internet giants raced to apply these new mobile tools to facets of urban life like transport, dining, services, entertainment, and more. This reshaped China’s urban landscape.
Guo Hong pioneered the concept of an “Avenue of the Entrepreneurs” in Beijing’s Zhongguancun district, using government subsidies and infrastructure upgrades to attract tech startups. This model was then scaled up nationally after Premier Li Keqiang endorsed “mass entrepreneurship and mass innovation.”
The central government directed local officials across China to establish thousands of new incubators, innovation zones, and government-backed venture capital funds. This flood of subsidies made it much easier for Chinese startups to get funding and work space.
Private venture capital in China exploded, quadrupling from $3 billion in 2013 to $12 billion in 2014. Entrepreneurship became a much more viable path, fueling a startup boom.
Beyond just money, the campaign shifted cultural attitudes to be more supportive of tech entrepreneurship. Parents who previously pushed their kids toward stable state jobs now encouraged startup dreams.
The inefficient but brute force approach allowed China to rapidly build a thriving tech ecosystem to rival Silicon Valley, catalyzing the growth of tech giants like Alibaba.
Chinese culture traditionally discouraged entrepreneurship, seeing it as risky. Internet entrepreneurship was unclear until the government endorsed it in 2014.
Jack Ma and Alibaba’s success made internet entrepreneurship seem achievable for ordinary Chinese people. Ma became a relatable hero figure.
Government endorsement and Ma’s example won over skeptical groups like mothers who had opposed entrepreneurship.
By 2015, Chinese were eager to join startups, flocking to firms like Sinovation. An entrepreneurship craze had gripped China.
Chinese internet companies dove into “online-to-offline” (O2O) services, using the internet to enable real world services like food delivery.
O2O took off rapidly in China’s crowded, polluted cities where people wanted deliveries rather than going outside.
Companies competed fiercely, subsidizing growth. The surviving firms like Meituan and Didi grew to huge valuations, transforming urban services.
WeChat’s ubiquity powered the O2O revolution by connecting users to services and enabling frictionless payments.
WeChat in China became a “super app” that allowed users to hail taxis, order food, book hotels, manage bills, buy flights, and more without leaving the app. This contrasted with the “app constellation” model in Silicon Valley where each app sticks to specific functions.
Chinese internet companies tend to take a “heavy” approach when entering new industries, wanting to control the entire process from end to end. In contrast, American internet companies take a “light” approach, focusing on their core competency of sharing information and connecting people digitally.
The “heavy” approach allowed Chinese companies like Meituan Dianping to scale up food delivery by hiring delivery fleets and subsidizing the process, gaining valuable real-world data. The “light” approach of Yelp failed to incentivize restaurants.
Mobile payments took off in China as Alipay and WeChat enabled payments by scanning QR codes. This frictionless system extended usage far beyond credit cards, allowing peer-to-peer transfers and micropayments. The resulting data gave Chinese companies a further edge.

The passage discusses how mobile payments and shared bikes took off rapidly in China, enabled by tech companies like Alibaba and Tencent pushing aggressive subsidies and promotions. Key points:

Mobile payments through apps like Alipay and WeChat Pay spread extremely quickly in China, facilitated by aggressive subsidies from Alibaba and Tencent. This leapfrogged credit cards which were not widely used.
Shared bikes like Mobike became ubiquitous in cities within a couple years, also enabled by mobile payments and subsidies. This created a reversal from bikes being seen as old-fashioned back to mainstream.
The data generated from mobile payments and shared bikes is massive and will help power AI development in China across many sectors like retail and real estate.
Overall, these innovations show how China’s tech ecosystem is able to rapidly solve practical problems by blurring online and offline worlds, in contrast to the more cautious approach of U.S. tech giants. The data generated will be a key asset for China’s AI development.
In 1999, China lagged far behind the U.S. in AI research. Chinese students had limited access to knowledge, relying on outdated textbooks and the occasional visiting lecturer.
Today, China has largely closed the gap with the U.S. in AI expertise. Chinese students now absorb the latest AI research in real time by studying online publications, joining WeChat groups, and watching lectures.
China is training an “army” of well-trained AI engineers who can team up with entrepreneurs to build practical AI products. The U.S. still leads in superstar AI researchers.
Seven corporate giants, split between the U.S. and China, are dominating the AI landscape. They are gobbling up talent and building private “power grids” to distribute AI across the economy.
The U.S. and China differ politically: China’s government actively supports AI growth through funding and incentives, while the U.S. system makes deployment slower.
Ambitious Chinese mayors are racing to make their cities AI showcases, optimizing traffic and installing facial recognition. China’s techno-utilitarian culture will enable faster AI adoption.
AI implementation relies more on an “army of tinkerers” applying existing breakthroughs, rather than a few elite researchers making fundamental discoveries. China is strong in training these tinkerers.
AI research is very open, with rapid sharing of algorithms, data, and results. This allows China to quickly absorb cutting-edge work. Chinese researchers actively discuss new AI papers and lectures.
China is now also contributing back to the global AI research ecosystem at an accelerating rate. Their growing participation led a top AI conference to reschedule to avoid conflicting with Chinese New Year.
The US has historically attracted the top AI talent through research funding and open academic environments. But recently some researchers have left due to visa issues or lucrative opportunities in China.
China is recruiting top talent through big salaries, research institutes like Baidu Research, and appealing to patriotism. The closed nature of China’s society may also appeal to some researchers.
Both the US and China have strengths in AI talent and research. But China’s top-down approach may accelerate implementation, generating more data and applications to fuel further AI progress.
Chinese AI research has grown rapidly, with Chinese authors nearly doubling their share of top AI papers from 2006-2015. Chinese institutions now rank highly in AI citation metrics.
Chinese researchers have made major contributions to neural networks and computer vision, including Microsoft Asia researchers developing the influential ResNet algorithm. Many top researchers have left Microsoft to found Chinese AI startups.
The “Seven Giants” of AI research (Google, Facebook, Amazon, Microsoft, Baidu, Alibaba, Tencent) conduct more proprietary research compared to academia. A major breakthrough at one of them could disrupt the open AI ecosystem.
Google/Alphabet is the clear leader among the Giants in AI investment and talent acquisition. Half of the top 100 AI researchers work at Google.
Baidu was an early mover in deep learning research, hiring Andrew Ng in 2014. But Google remains dominant, with unmatched resources devoted to AI R&D.
The next major breakthrough in AI is more likely to come from academia’s open research system than from the corporate giants focused on exploiting deep learning. But Google has the best odds of making a major discovery among the Giants.

Here are the key points from the summary:

Tencent and Alibaba are using their vast data resources to attract top AI talent and establish research labs globally, poaching researchers from companies like Microsoft.
The “grid” approach by giants like Google, Alibaba, and Amazon aims to commoditize AI by offering it as a standardized cloud service. In contrast, AI startups are taking a “battery” approach of building specialized AI products for specific use cases and industries.
There is an intense competition to develop AI chips optimized for machine learning. Silicon Valley leads for now, but China is investing heavily with the goal of surpassing the US.
In 2016, the Obama White House released a plan for harnessing AI in the US economy. But it had little impact compared to China’s national AI plan released in 2017, which outlined an aggressive strategy with billions in funding.
AI is both an economic and national security issue. How the competition between the “grid” and “battery” approaches shakes out, and whether the US or China leads in AI, will shape the new economic landscape.
In 2016, the Obama administration released a report on AI that generated little national response or new investment, in contrast to China’s AI plan released in 2017 which sparked major government funding and a national mobilization around AI.
China’s plan aims for the country to reach top tier status in AI by 2020, achieve major breakthroughs by 2025, and become the global AI leader by 2030. This catalyzed competition between Chinese cities to attract AI companies through massive subsidies, investments, and incentives.
The Chinese approach allows for some inefficiency but overall is extremely effective at rapidly deploying resources to upgrade the economy and technology. In contrast, the U.S. political climate harshly punishes failed government investments, making politicians wary of bold bets on emerging tech.
With autonomous vehicles, China’s techno-utilitarian approach will likely accept some downsides in lives lost for the greater good of massive reductions in traffic fatalities and efficiency gains. In the U.S., interest groups and ethical concerns around self-driving cars may slow deployment.
The AI revolution will come in four waves: internet AI, business AI, perception AI, and autonomous AI.
Internet AI and business AI are already here, reshaping the digital and financial worlds. They are driven by algorithms that learn our preferences and make recommendations.
Perception AI will digitize the physical world by learning to recognize faces, understand requests, and “see” the surroundings. It will blur the line between digital and physical.
Autonomous AI with self-driving cars, drones, and robots will have the deepest impact on our lives.
China is poised to lead in internet AI and perception AI, while the U.S. leads in business AI. The race is on for autonomous AI.
AI services will spread globally, with U.S. and Chinese companies competing fiercely for developing markets. China is investing in local startups while U.S. firms push their own products.
Each wave feeds off different data. Collecting and labeling that data will determine leadership in each wave. China excels at labeling with its large workforce.
First-wave AI has powered the rise of internet companies by optimizing online services and improving recommendations. Chinese companies like Toutiao are leading in using AI for content curation, creation, and monetization.
Second-wave business AI helps traditional companies optimize operations by finding correlations in their existing data. It outperforms humans at analyzing complex data to improve outcomes.
Companies like Palantir and Element AI provide business AI consulting services, with early applications concentrated in finance.
The U.S. currently leads in business AI due to large structured datasets and enterprise software at major corporations. But China is catching up rapidly as its tech giants also accumulate massive datasets.
In 5 years China may overtake the U.S. in business AI, aided by greater data availability, faster adoption by its companies, and a more flexible regulatory environment compared to the West.
Third-wave perception AI will power autonomous systems like self-driving cars and allow robots to interact naturally with humans. U.S. tech companies currently dominate.
China’s ecosystem of sensors, actuators, and reasonable regulation will give it an edge in adopting and benefitting from perception AI.
American companies have embraced enterprise software and standardized data storage, making it easier to apply AI for cost savings and profit maximization. This is not the case in China, where companies use idiosyncratic systems that are difficult to integrate with AI.
However, in some industries where AI can leapfrog legacy systems, China is making progress. For example, AI-powered microfinance apps like Smart Finance are expanding access to credit by using deep learning algorithms to analyze phone usage data to determine creditworthiness.
AI is also being applied to medicine and law in China to disseminate expertise more evenly. Startups like RXThinking are building AI diagnosis tools that empower doctors with medical knowledge from millions of cases. Likewise, iFlyTek is piloting AI tools that help judges cross-reference evidence and recommend consistent sentencing based on past cases.
The US currently leads in business AI overall, but China is making strides in certain leapfrog applications like microlending, AI-assisted medicine, and law. Key factors will be whether China can improve its data practices and integrate AI into traditional companies burdened by legacy systems.
The United States currently leads in business AI, with a strong advantage in optimizing processes and maximizing profits in industries like banking and insurance that have lots of structured data.
China may lead in applying AI to public services and industries that can leapfrog outdated systems, like healthcare and consumer credit. Its immature systems give it incentive to reimagine them from scratch.
Perception AI is the third wave, allowing machines to digitize the physical world through sensors, cameras, etc. This blurs the line between online and offline into “OMO” environments.
An example OMO environment is a supermarket where your cart recognizes you, suggests purchases based on your habits, guides you through the store, and facilitates purchase through facial recognition. Sensors track your cart’s contents precisely.
In summary, the US leads in business AI now, but China may lead in disruptively reimagining public services. Perception AI will create blended online-offline worlds, with early examples like facial-recognition payment and intelligent shopping carts.
AI-powered shopping experiences will feel revolutionary yet ordinary. They will eliminate friction points and tailor services to individuals while still following everyday patterns.
AI can tailor learning to each student and free up teachers. It can track comprehension, customize homework, automate grading, and connect students to remote tutors. China may leapfrog the US in education AI due to demand.
OMO requires collecting data from the real world. This includes video feeds, facial recognition, and voice commands.
Chinese citizens are more accepting of surveillance and data collection in public spaces in exchange for convenience. However, China has implemented some data protections.
There is an ongoing debate about balancing privacy and public data. China’s openness is giving it a head start on perception AI and digitizing environments.
The four waves of AI are internet AI, business AI, perception AI, and autonomous AI.
Internet AI involves machine learning algorithms applied to digital data like search, e-commerce, and social media. Business AI applies AI to enterprise operations and complex business data.
Perception AI gives machines sensory capabilities through hardware like cameras and sensors, allowing integration of physical and digital worlds. It is enabled by manufacturing hubs like Shenzhen.
Autonomous AI integrates the techniques of the previous three waves to enable machines to optimally navigate and shape physical environments. Early applications are in structured environments like factories, warehouses, and farms.
Chinese manufacturers have an edge in building hardware for perception AI due to the ecosystem in Shenzhen. This could give China a lead in implementing these technologies.
The four waves build on each other, with internet AI enabling business AI, and perception AI laying the groundwork for autonomous AI. Together they represent an progression in AI’s capabilities and applications.
Autonomous robots and AI are emerging first in commercial settings because they create direct economic value for companies.
Domestic robots are unlikely in homes soon, as tasks like cleaning and babysitting are still too complex for current AI.
Swarms of autonomous drones will enable new capabilities like house painting, firefighting, search-and-rescue, etc. China is likely to lead in drone swarm technology.
Autonomous vehicles will transform roads, cities, jobs. Companies like Google and Tesla have different approaches to developing self-driving cars.
Google is taking a cautious, perfectionist approach while Tesla is more incremental, deploying tech faster despite some risks.
China is likely to take a “Tesla” approach - deploying autonomous vehicles in controlled settings as soon as they’re better than humans, accelerating data collection.
China is building new cities and infrastructure tailored for autonomous vehicles, unlike the US which adapts them to existing roads. This will further accelerate China’s progress.

The article discusses the emerging competition between the United States and China in artificial intelligence (AI). It describes four “waves” of AI technology: internet AI, business AI, perception AI, and autonomous AI.

Currently, the United States leads in internet AI, exemplified by companies like Google and Facebook. However, China is making rapid progress and may catch up in 5 years. For business AI, China currently leads, with companies like Alibaba excelling in e-commerce AI. China is also ahead in hardware-intensive perception AI, with ambitious government programs like “Skynet” for surveillance. The US retains an advantage in software-based perception AI.

For autonomous AI like self-driving cars, the US currently leads but China is catching up rapidly. Key factors are whether the bottleneck is technical or regulatory, and whether Chinese-backed startups can outcompete Silicon Valley giants globally. The strategies differ, with the US conquering markets directly while China arms local startups.

Overall, China looks poised to take leadership in some critical waves like business and perception AI. This could tilt the geopolitical landscape and economic benefits toward China. But the AI future remains uncertain, contingent on factors like technological bottlenecks and successful globalization strategies.

The age of AI implementation has fueled speculation about artificial general intelligence (AGI) and the “singularity” where machines become superintelligent. This has divided thinkers into utopians and dystopians.
Utopians see AGI enabling human immortality and solving humanity’s biggest problems. Dystopians worry superintelligent AI could wipe out humanity while pursuing its goals.
However, AGI requires major scientific breakthroughs beyond today’s AI capabilities. Despite rapid recent progress, we are likely still decades or centuries away from human-level AI.
The true near-term crisis from AI will be economic and political, as current AI threatens widespread job loss and inequality within countries.
This crisis lacks the drama of AGI scenarios but could still deeply disrupt societies. We must address the impact AI will have on jobs, skills, and inequality.
The author believes AI has the potential to greatly widen economic inequality and lead to widespread technological unemployment. He sees parallels to the dystopian vision in Hao Jingfang’s sci-fi story “Folding Beijing.”
Massive productivity gains from AI will eliminate many jobs across industries and skill levels. This will hit developing countries hard by removing the low-cost manufacturing work that has historically kickstarted economic growth.
Within developed countries, AI will increase monopoly power of tech giants, concentrating wealth. This will worsen existing economic inequality.
The author argues this is different than past technological change. AI is a general purpose technology with the power to fundamentally disrupt the economy and society.
Techno-optimists dismiss these concerns as a “Luddite fallacy,” believing job losses will smooth out in the long run as with past innovations.
But the author argues AI’s scale sets it apart. We can no longer rely on blind optimism about technology’s impact on jobs and inequality.
McAfee described general purpose technologies (GPTs) like the steam engine, electricity, and information technology as the most important innovations that accelerate economic progress.
Looking only at GPTs provides limited data points for evaluating technological change and job losses.
Historically, GPTs like the steam engine and electricity enabled low-skilled workers to be more productive in factories, increasing overall prosperity.
However, the most recent GPT, information and communication technology (ICT), has coincided with wage stagnation and rising inequality, unlike past GPTs.
ICT is more skill-biased, benefiting high-skilled workers disproportionately. This contributes to economic stratification.
AI will be the next GPT, bringing an economic revolution larger and faster than prior ones. PwC predicts AI will add $15.7 trillion to the global economy by 2030.
AI will automate both physical and cognitive tasks. It will replace human jobs rather than facilitate deskilling.
The AI revolution will happen faster than past GPT-driven transformations because AI algorithms can be instantly distributed and improved digitally.
AI is a general purpose technology (GPT) that will likely have a major economic impact, like previous GPTs such as steam engines, electricity, and information technology.
Unlike past GPTs, AI’s skill biases suggest it may lead to negative effects on employment and income distribution.
AI excels at narrow, data-driven tasks but struggles with dexterity, creativity, and complex strategy. This creates winners and losers in job replacement.
Jobs are categorized into four risk quadrants: high risk of replacement (“Danger Zone”), low risk (“Safe Zone”), moderate risk where machines do computational work and humans provide social interface (“Human Veneer”), and moderate risk where automation will slowly encroach on human capabilities (“Slow Creep”).
Estimates of job loss from automation range widely, from around 50% of jobs being at high risk of automation to less alarming figures below 10%. The impact will depend on rate of AI advancement and adaptation by companies.
China’s contribution to AI development and application will accelerate impacts. Venture capital funding and digital dissemination of AI tools are also key accelerants.
While past GPTs boosted productivity and jobs, AI’s biases suggest negative impacts on employment are likely. The scale depends on pace of advancement and adaptation.
Economists have produced widely varying estimates of the impact of automation on jobs, ranging from 9% to 47% of jobs being at high risk.
Earlier studies took an “occupation-based” approach, looking at entire occupations. Later studies took a more granular “task-based” approach, breaking jobs down into specific tasks.
Task-based studies tend to find lower risks of automation compared to occupation-based studies. The OECD study found only 9% of US jobs at high risk, while the PwC study found 38% at high risk.
The author believes the actual risk is higher than the low-end estimates like the OECD’s because those were based on expert predictions made in 2013 that underestimated the rapid improvements in AI since then.
Beyond one-to-one replacement of human workers, the author argues that AI will also enable new business models that could disrupt entire industries and their workforces from the ground up.
The task-based approach focuses only on automating existing jobs and misses these potential industry-wide disruptions enabled by AI. The author believes this could put even more jobs at risk beyond the 38% estimated by PwC.
Automation enabled by AI will likely displace 40-50% of jobs in the US over the next 10-20 years. This includes both one-to-one replacements (38% of jobs) as well as disruption from the ground up (10% of jobs).
Factoring in new job creation and inertia, net unemployment could be 20-25%. A 2018 Bain study reached a similar conclusion, predicting employers will need 20-25% fewer employees globally by 2030.
Contrary to conventional wisdom, China may be less impacted than the US in the near term. This is due to Moravec’s Paradox - it is easier to automate intellectual tasks with AI than physical ones with robotics.
White collar jobs are more immediately threatened by AI algorithms which are adept at thinking. Blue collar jobs depend more on robotic dexterity which lags behind human capability.
In the US, AI algorithms stand to white collar workers as tractors were to farmhands - greatly displacing their labor through productivity gains. China has more time before robotic automation fully eliminates its manufacturing jobs.
AI will widen the gap between the AI “superpowers” (the U.S. and China) and the rest of the world. The superpowers will capture most of the economic gains from AI, while other countries will be left behind.
AI has a natural tendency towards monopolies due to network effects and accumulating data advantages. This could lead to winner-take-all economics in many industries.
The superpowers will see rising inequality as corporate profits surge but many middle-class jobs are automated away. This will bifurcate the labor market into highly-paid and low-paid work.
Developing countries will lose their low-wage manufacturing advantage. With factories and services automated, they may become dependent on the AI superpowers.
Within the superpowers, the “great decoupling” of productivity and wages will accelerate. AI monopolies will drive corporate concentration and profit growth but eliminate many middle-class jobs.
The optimistic vision of AI democratizing access and empowering people worldwide is unlikely to transpire. Without intervention, AI is more likely to exacerbate inequality both globally and domestically.
Kai-Fu Lee had lived his life obsessed with work and driven by algorithms to maximize influence, neglecting time with family. This was challenged when he was diagnosed with cancer in 2013, forcing him to confront his mortality.
This led him to reshuffle his priorities, spending more time with family and being more empathetic. He realized maximizing influence did not bring meaning, but sharing love with those around him did.
His cancer went into remission, but the experience gave him a new vision for how humans can coexist with AI. While AI will create economic value and destroy jobs, we should not equate our self-worth with economic value.
Instead of trying to survive in the age of AI, we have an opportunity to focus on what makes us human - things like love and connections. We can use AI to free us to build these emotional connections.
His own life shifted from optimizing for influence back to focusing on people and love after his confrontation with mortality. The book explores this path further.
Kai-Fu Lee was obsessed with work and maximizing productivity, viewing all parts of life like variables in an algorithm. He nicknamed himself “Ironman” and worked constantly, even after surgeries.
Lee gave speeches urging young Chinese people to work hard and make a mark during China’s rapid growth. He showed a tombstone reading “Here lies Kai-Fu Lee, scientist and business executive” who helped turn tech advances into products.
Later, as a mentor, Lee changed his fictional epitaph to read “Here lies Kai-Fu Lee, who had a love for education” and helped many students who called him “Teacher Kai-Fu.”
Going for a medical checkup, Lee was told he needed a PET scan after initial scans found something. The PET scan starkly revealed he had late-stage cancer throughout his body.
The diagnosis shocked Lee and forced him to confront his mortality and reevaluate his life’s priorities. He realized his tombstone epitaphs were foolish - he had maximized influence but not meaning.
After receiving a cancer diagnosis, Lee struggled to handwrite his will in traditional Chinese characters, which he had not regularly written in decades. This task forced him to confront his mortality.
Viewing his PET scan showing many cancerous tumors threw Lee into despair and self-pity over dying young despite his achievements.
Writing down his wife’s and daughters’ names refocused Lee on the importance of generously sharing love with family, which he regretted not doing enough while focused on career success.
With his mother in declining health nearby, Lee was filled with remorse over not expressing his love for his parents before it was too late.
Facing death made Lee realize the importance of sharing love with others over individual success and achievement. He regretted living so self-centeredly without giving love.
Lee was diagnosed with late-stage cancer, causing him to reflect deeply on his priorities and regrets. He realized he had not focused enough on relationships and love.
A Buddhist monk, Master Hsing Yun, bluntly asked Lee what his goal in life was. When Lee replied “to maximize my impact and change the world,” the monk cautioned him that this mindset often disguises ego and vanity.
Master Hsing Yun advised Lee that constantly calculating and quantifying everything suffocates love, which is the one thing that truly gives life meaning. Humility and recognizing our smallness is key to sharing love with others.
Lee came to understand how destructive his old algorithmic way of thinking had been, and struggled to replace it with a new way of being human that didn’t mimic that computational approach.
While getting a second opinion on his cancer diagnosis and prognosis, Lee continued researching the disease extensively. He realized the staging system, while straightforward, did not capture subtleties affecting outcomes.
His story illustrates a journey from a relentless focus on optimization, impact and achievement to prioritizing human relationships and love. Master Hsing Yun’s words prompted deep reflection on living these values rather than just understanding them intellectually.
The author was diagnosed with stage IV lymphoma based on the traditional staging system, which categorizes the cancer based on simple factors like number and location of tumors. This gave him an estimated 50% 5-year survival rate.
He later found a more data-driven study that incorporated additional prognostic factors beyond just tumor spread, like patient age, tumor size, bone marrow involvement, etc. Based on this rubric, his prognosis improved to an 89% 5-year survival rate.
This demonstrated to the author the limitations of the traditional staging system which is based on simple factors for memorization, versus modern data-driven medicine.
After recovering from cancer, the author vowed to cherish his time with loved ones more, avoiding optimizing his schedule and priotizing people equally regardless of their status or talents.
His cure had two pillars - the medical technology/data-driven diagnosis, and the emotional healing from connecting with family and friends. Both were essential for saving his life and changing how he views human relationships.
AI algorithms will increasingly perform diagnostic and prescriptive functions better than individual humans can. Doctors may use AI as a tool or be replaced by it entirely in some cases.
However, AI lacks the ability to provide the love, empathy, and human connection that is most needed for human flourishing.
Humans uniquely have the capacity to love and be loved. Love is what makes life meaningful.
We must build a future that combines AI’s analytical abilities with humanity’s essence of love. This requires reimagining and restructuring society.
If we unite behind fostering love and compassion, humans can thrive in the age of AI, attaining both material prosperity through AI and spiritual flourishing through human relationships.
Mass unemployment and inequality are imminent with rapid AI development. This risks instability if left unaddressed.
We must proactively reconstruct economies and rewrite social contracts to distribute the material abundance of AI while promoting human compassion.
With vision and unity behind enhancing human love alongside AI capabilities, we can create a society that enables unprecedented human flourishing.
The transition from an industrial economy to an AI economy will require not just economic changes, but a shift in culture and values away from defining ourselves by our jobs.
Purely technocratic solutions like universal basic income don’t address the human need for purpose and community. Instead, we should use AI’s economic bounty to support activities that make us more human.
Proposed “technical fixes” from Silicon Valley fall into three categories: retraining workers, reducing work hours, and redistributing income. Each has limitations in fully addressing massive job displacement.
Chinese technologists are optimistic that AI will create prosperity like previous technological advances. The government is also expected to help displaced workers transition. However, the challenges may be underestimated.
More creative solutions are needed beyond these technical fixes, involving private sector job creation, investment strategies, and policy. The goal should be increasing human prosperity and flourishing, not just preventing economic hardship.
A multifaceted approach and bold experimentation will be required to find a smooth transition to an AI economy. The potential rewards make it worth the effort to get this social transformation right.
Education and retraining will not be enough to address the widespread labor disruptions caused by AI. More radical redistributive measures like universal basic income (UBI) are gaining interest.
UBI proposals involve giving every citizen a regular stipend with no strings attached. Funding would come from taxes on major tech companies and winners of the AI revolution.
Silicon Valley elites see UBI as a solution to technological unemployment and a way to avoid unrest. But their interest may also be self-interested - UBI could protect them from public backlash.
While UBI seems like a quick fix, we should think critically about its motivations and effects before embracing it as the solution. Will it truly benefit all, or just pacify the losers of automation while benefiting the tech elite?
Overall, more creative solutions are needed that address underlying inequities and ensure technology spreads prosperity. UBI alone is likely not the magic wand solution that Silicon Valley hopes for.
Many in Silicon Valley see universal basic income (UBI) as a “magic wand” that can make the negative impacts of AI disappear without requiring tech companies to change their behavior or business models.
UBI allows tech companies to avoid taking responsibility for the deeper psychological and social impacts of automation. It is a simplistic, digital-only solution that doesn’t address real-world complexities.
Rather than settle for UBI as an economic floor, we should view AI as an opportunity to create more humane jobs that emphasize uniquely human skills like compassion.
The private sector should take the lead in developing new human-AI symbiotic jobs where AI handles optimization but humans provide the creative, compassionate touch. This is already happening in medicine with AI diagnosticians and human caregivers.
Sharing economies and platforms will enable more customized service jobs to flourish. Entirely new humane service roles we can’t envision today are likely to emerge.
To fully leverage AI’s benefits, those in professional roles should embrace rather than resist AI tools. Symbiosis will be rewarded over the long-term.
The goal should be using AI to inch society toward greater kindness and love rather than settling for UBI as a way for tech companies to absolve themselves of responsibility.
The author believes that while profit-seeking businesses may create some new human-centered jobs in the AI economy, market forces alone will not be enough to offset massive job losses and inequality.
The author sees a need to reinvigorate corporate social responsibility, impact investing, and social entrepreneurship to encourage and reward prosocial activities and humanistic service jobs. This includes a new form of impact investing focused on scaling up human-centered services that provide meaningful jobs.
Orchestrating a broader cultural shift will require going beyond private sector initiatives. Public policies and government action will be needed to fundamentally change economic structures and incentivize humanistic service jobs on a large scale.
The author cautions against relying solely on universal basic income, which he sees as too individualistic and divisive. He advocates instead for policies that strengthen social bonds and community engagement.
Overall, the author argues that responding to AI’s impacts will require a multifaceted approach spanning the private sector, impact investing, and significant government intervention to support mass numbers of humanistic service jobs. This is necessary to build an AI economy centered on compassion and human connections rather than increasing inequality.
The author was inspired by volunteers he saw at a Buddhist monastery in Taiwan, who contributed their time to help others find serenity. This made him think about how to build a more compassionate society.
Rather than a universal basic income (UBI) that pays people for nothing, the author proposes a “social investment stipend” - a government salary for those who invest time in care work, community service, and education.
This would create an army of people to help the needy, perform services like environmental cleanup, and allow people to pursue training or hobbies. It would foster a culture of compassion.
Requiring social contribution to receive the stipend reinforces that economic abundance should be used to recommit to one another, not just numb the pain of job losses.
There are open questions around amount of stipend, assessing performance, and which activities qualify. Administering this on a large scale could be complex.
But the author believes this is better than UBI’s laissez-faire individualism, and would nudge society in a more human direction by using AI’s economic bounty to strengthen social bonds.

Here is a summary of the key points regarding avoiding an “AI race” mentality between the US and China:

Framing AI development as a “race” is unproductive and zero-sum. It pits China’s gains as America’s losses.
This mentality has led some in the US to use China’s AI progress to spur action, arguing the US risks losing its edge in military AI applications.
But AI development should not be viewed as a Cold War-style arms race. AI has huge potential for shared progress and prosperity.
The US and China should collaborate where possible, while competing in narrow domains like defense. Healthy competition in civilian AI can benefit both countries.
Avoiding an “AI race” mentality will allow the US and China to shape a shared AI future focused on human flourishing rather than just geopolitical dominance.
The author’s experience bridging the US and China gives him optimism that mutual understanding and shared AI progress is possible between the two countries.
The current AI boom is more akin to the Industrial Revolution than an arms race. AI’s greatest potential for disruption lies in labor markets and social systems, not military contests between nations.
To responsibly guide AI’s development, we need diverse global wisdom on reforms to education, cultural values, notions of progress, privacy, and governance. Examples include South Korea’s gifted education, America’s social-emotional learning, Switzerland’s craftsmanship culture, Canada’s volunteering ethic, China’s intergenerational families, and Bhutan’s Gross National Happiness.
Humans must maintain agency in shaping AI’s future based on our values. Dystopian outcomes are not inevitable if we focus technology on human flourishing beyond economic optimization.
The key to understanding ourselves lies not in mimicking cognition but in cultivating love, family, and community. AI should free us to focus on our hearts, not just optimize our brains.
By coming together across borders and classes, we can write an uplifting story for AI based on using machines for productivity while loving one another. We must choose to let machines be machines and let humans be humans.
The author thanks various people for their help and support in writing the book, including Rick, Erik Brynjolfsson, James Manyika, Jonathan Woetzel, Paul Triolo, Shaolan Hsueh, Chen Xu, Ma Xiaohong, Lin Qi-ling, Wu Zhuohao, Michael Chui, Yuan Li, Cathy Yang, Anita Huang, Maggie Tsai, and Laurie Erlam.
He is especially grateful to his colleague Rick for pushing him to be the best he could be.
He also thanks his family for putting up with his distraction while writing the book over the past six months, and says this should be his last book for a while, though he has told them that seven times before.
The main points are expressing gratitude to all those who helped with the book, particularly colleagues, co-authors, and family members for their support, feedback, and patience.

Here is a summary of the key points from the sources you provided:

Frederick Jelinek argues that statistical models and machine learning are superior to traditional rule-based approaches for natural language processing tasks like speech recognition and machine translation. He believes linguists should embrace these techniques.

The news app Toutiao has become extremely popular in China, with users spending an average of 74 minutes per day on the app. It uses AI to analyze user behavior and customize content recommendations.

A “new standard of beauty” driven by AI facial recognition is leading many Chinese women to undergo plastic surgery to look more “beautiful” according to the AI.

In Jiangsu province, an “AI prosecutor” system helps rank prosecutors and evaluate their work. This kind of AI monitoring of human performance is spreading in China.

By the end of 2017, Baidu had sold over 85 million Xiaodu smart home devices with integrated AI voice assistants.

Drone maker DJI dominates the consumer drone market, with over 70% global market share in 2017. Its advanced AI and robotics give it a strong competitive advantage.

Google’s self-driving car program had accumulated 1.5 million miles of autonomous driving as of 2016, compared to Tesla’s 47 million miles driven by Autopilot.

China plans to invest $583 billion into the Xiong’an New Area to create an AI and tech hub.

Many predict advanced AI or the “singularity” will arrive by 2045 or sooner, raising concerns about existential risks from AI. But the real crisis is how AI will transform the economy and jobs.

While AI automation could displace many jobs, historical industrial revolutions suggest new jobs will also be created. The challenge is how to manage the transition and ensure broadly shared prosperity.

In 2017, AI startups received a record $15.2 billion in funding, up from $5 billion in 2016, but far short of the amount needed for academic research and retraining programs. Avoiding an angry populist backlash will require creative policy solutions to reduce inequality, like universal basic income.

The key is to focus not just on AI’s economic impacts, but also on how AI can improve education, healthcare, and quality of life if steered wisely for the common good. Doing so will require compassion and wisdom.

Here is a summary of the key points from the Business Insider article about Mark Zuckerberg’s Harvard commencement speech:

In his speech, Zuckerberg said that exploring a universal basic income will become increasingly important as more jobs become automated by AI and robots.
He argued that a basic income would provide people with a “cushion” as they transition between jobs and careers.
Zuckerberg also talked about the importance of purpose, community, and building projects together. He said finding purpose is key to building community.
He cited universal basic income as one example of how we can work together to build an economy that works for everyone, not just the wealthy and powerful.
Zuckerberg’s comments signal a shift in thinking among tech leaders about dealing with the potentially disruptive effects of AI and automation on employment and incomes.

The article also notes that two of the fastest growing professions are home health aides and personal care aides, which pay just over $20,000 per year on average. This highlights the issue of low wages in some growing fields, which proponents argue a basic income could help address.

Here are the key points summarized from the passages on machine learning:

Machine learning, especially deep learning, has driven recent advances in AI and the economy. It allows computers to find patterns and make predictions from large datasets.
Deep learning is a subset of machine learning based on neural networks modeled after the human brain. It has led to breakthroughs in computer vision, speech recognition, and other fields.
The availability of big data, faster computers, and algorithmic advances have accelerated progress in deep learning over the past decade. This has enabled practical applications across industries.
Machine learning and its advances in pattern recognition are behind breakthroughs like self-driving cars, medical diagnosis systems, and AlphaGo beating top human players.
As machine learning continues improving, it threatens certain jobs through automation but also may create new industries and opportunities. Managing this economic transition will be a major challenge.
Overall, machine learning represents a powerful general purpose technology that is transforming the economy, similar to past innovations like electrification. But its social impacts require thoughtful responses.

Here is a summary of the key points related to the universal basic income discussion in the book:

The idea of a universal basic income (UBI) has gained traction as a potential policy response to technological unemployment caused by AI and automation. It involves providing every citizen with a regular stipend to cover basic needs, regardless of income level or employment status.
Some Silicon Valley figures have endorsed UBI, arguing that it could provide economic security and stimulate entrepreneurship if jobs disappear due to automation. Critics claim it could discourage work and be prohibitively expensive.
The author discusses various pros and cons of UBI and evaluates its feasibility. Potential benefits include reducing poverty and inequality, supporting work transitions, and giving people more freedom and security. Concerns include cost, effects on motivation to work, and proper implementation.
The book suggests UBI may not be an ideal or sufficient solution on its own. Other policy responses should be considered as complements, like retraining programs, reducing work hours, employee profit sharing, and encouraging social entrepreneurship.
Overall the book argues UBI merits serious consideration and debate as a potential policy option, but many open questions remain regarding its viability and design. The impact of AI and automation on the labor force is still unfolding, so policymakers should continue examining UBI and alternative responses.
Tencent and its WeChat platform have been hugely influential in China’s internet landscape. WeChat combines messaging, social media, payments and more into one “super app”.
Chinese tech companies initially copied Western models like Facebook, but adapted them for the Chinese market. This allowed homegrown firms like Tencent to dominate.
China’s internet ecosystem developed separately from the West, with different platforms, norms and regulations. The “Great Firewall” blocks Western services like Google and Facebook.
AI development is progressing in four waves - internet AI, business AI, perception AI, and autonomous AI. China is racing ahead in consumer/business AI while the US leads in cutting edge AI research.
There are concerns about potential dystopian consequences of AI, but the technology also has promise to improve lives. Balanced policies and norms are needed.
Lee offers suggestions for constructive cooperation between the US and China on AI, and principles for managing AI’s impact on jobs, privacy and ethics. He advocates preparing for AI’s societal impacts.

#book-summary

AI Superpowers - Kai-Fu Lee