Two fears that dictate the future of AI
By Florence So
Currently, there exists only two emerging AI superpowers: United States and China. As Kai-Fu Lee points out in his latest book, the recipe for AI success requires four ingredients: AI expertise, government support, data, and entrepreneurial ecosystem. In his opinion, the fundamental cultural differences between the West and the East inform how these two superpowers will play out in the future. And as such, China will make significant progress in the next five to ten years. I agree with him, and further support his predictions by adding the element of fear. In politics all over the world, fear is a tactic to retain power. But in the AI space, I am talking about different kinds of fear. In the West, the fear is that of privacy invasion, of harm, and of disruption. In the East, the fear is that of failing the commands of the Chinese authoritarian state. Both fears contribute in their own ways toward the developments of the AI superpowers.
Machine behaviour is acquired with a large volume of data, without which learning is impossible. For example, we feed ridiculously large sets of bio data to the machine to find correlations between certain genome-sequencing and brain cancer. This approach sounds logical. High risk individuals can seek preventive help. But it is also murky. First, who owns the data? Who can use it, and for what purpose? Second, what if employers get a hold of this data and discriminate those with high risk at hire? What if your potential is limited by machine learning algorithms? Think “Gattaca”, the 1997 film.
In the West, the expectation of privacy is constitutionally guaranteed. The overall sentiment in America and Europe alike is the fear of losing privacy and freedom; or yielding too much control to private giant corporations like Google, Facebook, Apple, and Amazon. In response, policy makers debate over issues such as data sovereignty, cybersecurity, machine bias, and so on.
In China, privacy is a luxury. Privacy rights have been available to citizens under their Constitution since the 1980's. But in reality, what is written in the Constitution need not apply. For centuries, the Chinese people have been subjected to state control and surveillance. No wonder citizens don’t really mind trading ‘privacy’ for convenience. As a result, massive collection of online to offline (O2O) data has long begun. Nobody in China complains about their Social Credit System. In fact, they see it as a system to regulate trust. China has an enormous population and the cities are dense. The country is a gold data mine of consumption and transportation habits. While data is collected by voluntary upload of select users over here, unobstructed data collection of the lives of billions give China a competitive advantage, in quantity and granularity. The sheer volume coupled with O2O quality data makes any type of machine learning a piece of cake.
Trump’s politics of fear has manipulated many to buy in protectionism and other agendas. But people can still vote him out if they really want to. In China, ordinary citizens are stuck in a fear they can’t get out of: the fear of the State. There is no such thing as the rule of law in China, as the justice system is controlled by the Communist Party of China (CPC). Torture, forced confessions, arbitrary detention, and poor access to lawyers are notorious problems all over China. As such, most people dare not express their opinions, especially those that don’t align with the CPC. Nobody wants the “Great firewall of China” but few will speak up for the real internet. This fear extends to students and expats in foreign countries. Outside China, they believe they are somehow still being watched and are reluctant to share their true political views.
Besides speech, Chinese citizens must comply with actions. When the soft authoritarian, nationalistic, populist Xi Jinping announced that China will be the world leader of AI by 2030, everybody listened. This top-down approach forces every private company to align their mission with the Party’s collective one. While Google, Facebook, Apple and Amazon are driven solely by profits, Alibaba is backed by, and under tight control of, the Chinese government. The owner, Jack Ma, recently became an official member of the CPC.
In the West, smart city initiatives such as Sidewalk Toronto are run by private companies like Alphabet, the parent company of Google. In the East, smart cities are orchestrated by the State. China picks an empty city and moves AI start-ups, venture capitalists, and smart people in. Over here, self-driving cars must overcome the Trolley problem as well as Level 5 perfection before mass acceptance. Over there, the State build infrastructures to accommodate technology. Self-driving cars will stay on their own AV lanes. In any case, Trolley is not a big concern as a human’s life has little value in the grand scheme of the CPC. The Chinese government has always been techno-utilitarian; and therefore, when it comes to new technology, “Let’s get it out there and fix the problems later.”
Moreover, China has always done things their own way. They implemented their own GDPR. The Beijing AI Principles was issued by the Beijing Academy of Artificial Intelligence (BAAI), an organization backed by the Chinese Ministry of Science and Technology and the Beijing municipal government. Despite, China shows no signs of scaling back on its schemes for tracking and monitoring citizens. In fact, the law explicitly requires most online services operating in the mainland to surrender information to law enforcement.
In America, everyone must fend for themselves in the wake of job market disruption. Protest to government, strike, redistribute income (UBI), whatever it takes. Fear is in the air; whereas in China, AI is full speed forward due to annual decline in labour force. The manufacturing sector simply needs automation to survive. There is no concern for disruptive job replacement. Clearly, government orders as well as the labour climate are facilitating the march towards AI.
The East and the West have opposing attitudes towards machines and robots. In the U.S., Alexia calling 911 was considered invasion of privacy. Contrastingly, in Asia, humans marry robots. In addition, online merge offline (OMO) has been accepted fully in China. Your grocery cart greets you by name, reminds you what’s in the fridge, and suggest new products. You will pick up the necessary and walk out with a smile (i.e. pay by facial recognition). China has gone cashless a long time ago, skipping the entire credit payment system. Every transaction is either through WeChat or Alipay. There is no minimal dollar amount nor point of sale machines required. Even beggars have their own QR codes.
User adoption faces resistance in the U.S. because the concepts of IoT and OMO freak people out. Adoption is an uphill battle. Success is not credited yet every failure is scrutinized. For instance, human drivers get into accidents every single day, but one self-driving car accident will become the talk of the town. The ethicists hold up legislation allowing for adoption. Furthermore, AI experts fear AI abuse. Trust, algorithmic explicability, machine accountability, fairness are serious concerns, and for good reasons, while the Chinese political culture couldn’t care less about reaching a moral consensus on every ethical question.
AI academics embrace an open and speedy culture. For example, any scientific discovery is instantly published on arxiv.org. Although the elite research community resides in the U.S. and increasingly Canada, immediate impacts rest on mass implementation anywhere. While research is way ahead, and entrepreneurial projects are driven by ideals of free market in the West, China has an army of good-enough implementors backed by government subsidies and directives. An AI breakthrough comes every few decades, but new applications of the technology reap their fruits everyday and have no geographical limits. Again, this ecosystem allows the Chinese to progress quickly.
Not only is the entrepreneurial ecosystem in China healthy, inventors are solving local problems. With a population of nearly 1.4 billion, there are plenty of local problems. For instance, many people with low credit ratings are rejected by traditional lenders. So, microloans such as Smart Finance took off and has had very low default rate. The barrier of entry into AI deployment is low in China. Mom and pop shops can afford AI powered delivery systems. Airbnb-like apps can manage food delivery and cleaning services within the same platform. The government dedicates an entire street for venture capitalists, contrary to the U.S. who don’t believe in building Silicon Valley by bricks.
As we have seen, our fears have a big role in dictating our actions and the future of machine learning R&D. In the U.S., AI research is significantly more advanced, but adoption is hindered because people are concerned about AI invading privacy, taking over the job market, and doing harm. In China, the knowledge of AI is largely borrowed but massive adoption is unstoppable because the State says so and no one dares to do otherwise.
That’s where we are at in the U.S. and in China. Now, where does Canada fit in the spectrum? What are we afraid of? What does the future of CANADA.ai look like?
This content has been updated on 10/07/2019 at 14 h 31 min.