China has been booming for decades. The country has not only reduced extreme poverty from 88 percent to under 2 percent in just 30 years, but has also become the world's factory for high technology. Although the growth rate is slowing down slightly due to the aging population, China is still one of the major players in many technological fields. One of these areas, and perhaps the most significant, is the development of artificial intelligence. The Chinese government announced in 2017 its goal to become the world leader in artificial intelligence by 2030 and has since invested billions of euros in AI projects in research and business. The government's venture capital fund is investing over 30 billion euros in AI.
The northeastern city of Tianjin alone has allocated 16 billion euros for promoting AI, and a 2 billion euro AI research park is currently being built in Beijing. In addition to these enormous investments, the government and companies in China have access to an unprecedented amount of data, ranging from citizens' health to their smartphone usage. WeChat, a multifunctional app that allows users to chat, make appointments, navigate, transfer money, order rides, read news, and much more, grants the Communist Party full access to user data. Through WeChat alone, the Communist Party is able to create behavioral profiles of virtually everyone in the country, whether citizens or foreign visitors. And this is just one (albeit very large) source of data that the state can access. Additionally, there are millions of surveillance cameras and the like. Many believe that this wealth of data will make China a powerhouse in AI, surpassing the USA and Europe. However, AI involves more than just data, and progress requires more than just investing billions of dollars. Analyzing China's potential to become a global leader in AI - or in any other technology that requires consistent innovation - from various perspectives paints a sobering picture. During my research trip to China last year, I visited large companies and startups in Beijing, Shanghai, and Shenzhen (China's Silicon Valley) in search of tech innovations. To put it in one word: disappointment.
I was massively disappointed in terms of innovation. With the sole exception of the drone manufacturer DJI, I saw nothing that I hadn't seen elsewhere in the world before. When I asked why, it became clear to me that China has a systemic problem that has prevented outstanding innovation until today and will likely continue to do so. First of all, one must know how innovations come about. Many great inventions have arisen by chance, and some of the world's most successful companies began in garages, student dormitories, or under similarly inconspicuous circumstances (including Google, Facebook, Amazon, and Apple, to name a few). Innovations arise through serendipity and, as I describe in my book "Kill your agency" (Haufe, 2015), through the recombination of existing facts/events. When scientists at universities interact with inventors and entrepreneurs through the freedom of research and teaching guaranteed to them by the Basic Law (Article 5), great things happen. Observing China, one can see the huge number of intelligent people in the country and quickly conclude that with so much potential, it must be easy to develop innovations.
However, looking at its tech and innovation prowess, one is disappointed to find that China has historically built on technologies developed elsewhere but has not yet demonstrated a track record of innovation. Oxford scholars Carl Benedikt Frey and Michael Osborne pursued the same observation and found in their study published in Foreign Affairs in June 2020 that none of the top 100 most cited patents from 2003 to the present originated from China! The tech giants Tencent, Alibaba, and Baidu are all very successful in the Chinese market, but they have their roots in technologies or business models originating from the USA and adapted for the Chinese population. This is not an attempt to diminish the tech companies and their success, but often travelers to China like me and other reporters fall into the trap of confusing China's catching-up pace and incomparably large market size with innovation power. In other words, it is not a significant technical challenge to operate a train from Beijing to Shanghai at a speed of 350 on a perfectly straight track built for that purpose and then arrive on time. In contrast, it is an almost unbelievable feat to ensure that an ICE train between Hamburg and Stuttgart arrives on time when it must share the track with over a thousand S-Bahn trains, freight trains, regional trains, and, of course, seat-blocking demonstrators advocating for the preservation of the stag beetle. The Communist Party's primary task is to keep the huge country with its 22 provinces and 1.4 billion inhabitants united. And this can only be achieved with an iron fist or, as my Chinese tour guide Bruce Lee (that was really his name) explained to me, when taking an excursion with a kindergarten group, everyone must be closely guided with a firm hand, otherwise chaos ensues. While this may be correct from China's historical perspective, from the perspective of innovation research, we know that the most innovative societies have always been those that allowed people to pursue controversial ideas. China's strict internet censorship and citizen surveillance do not exactly encourage the pursuit of controversial ideas. The country's social credit system rewards those who follow the rules and punishes those who deviate.
It is almost impossible for a culture of social conformism to have a positive impact on technological innovations. An example where conformism went wrong is the Soviet Union, which, despite high investments in science and technology, briefly competed with the USA in areas such as nuclear energy and space exploration but ultimately lagged far behind, mainly due to political and cultural factors as well as the lack of genuine competition. China's significant advantages in capital, data, and talent are not necessarily advantageous if its boundaries of thought are restrictive and need to remain restricted to maintain the system. Even though billions are now being invested in the education system and students are being drilled in computer science - the realization remains - the best students are not necessarily the best researchers. To be a good researcher, one must question existing knowledge and also come up with new ideas. To get an idea of who is likely to take the lead in AI, it may be helpful to first consider how the technology will evolve beyond its current state. To put it plainly: AI is currently somewhat stuck. Algorithms and neural networks repeatedly achieve new and impressive feats - such as DeepMind's AlphaFold, which accurately predicts protein structures, or OpenAI's GPT-3, which writes blog articles based on short prompts. However, the capabilities of these systems are still largely defined as weak intelligence: performing a specific task for which the system was painstakingly trained using vast amounts of data. (It should be noted at this point that some have speculated that GPT-3 from OpenAI could be an exception, the first example of machine intelligence that, while not "general," has surpassed the definition of "weak"; the algorithm was trained to write text but eventually was able to translate between languages, write code, automatically complete images, perform mathematics, and execute other language-related tasks for which it was not specifically trained. However, all of GPT-3's abilities are confined to skills it learned in the language domain, whether spoken, written, or in a programming language). The success of both AlphaFold and GPT-3 was largely due to the huge datasets they were trained on; there were no revolutionary new training methods or architectures involved.
If all that would be needed to advance AI is a continuation or scaling of this paradigm - more input data results in higher performance - then China could indeed have an advantage. However, one of the biggest hurdles AI needs to overcome to evolve in leaps rather than small steps is precisely this dependence on extensive, task-specific data. While China's treasure trove of data currently gives the country an advantage, it may not necessarily be advantageous on the path to AI dominance. The abundant data is useful for developing relevant products today, but not for advancing the development of artificially intelligent systems. For example, WeChat data on users' spending habits is valuable for developing AI that helps people save or suggests articles they might want to buy. This already leads and will lead to highly customized products that will bring in a lot of money for their inventors and the companies that sell them. However, the volume of data is not what will drive AI forward. As Frey and Osborne put it, "data efficiency is the holy grail for further advances in artificial intelligence." To this end, research teams in academia and the private sector are working on ways to make AI less data-hungry. New training methods such as One-Shot Learning and Less-than-One-Shot Learning have already emerged, along with countless efforts to make AI learn more like the human brain. Although these advances are not insignificant, they still fall into the category of "baby steps". No one knows how AI will evolve beyond these small steps - and this uncertainty, according to Frey and Osborne, is a major impediment to China's rapid path to AI dominance. Beyond the question of whether China will achieve dominance in AI, is the question of how it will use the powerful technology. Some of the methods China is already using AI for could be considered morally questionable, from facial recognition systems aggressively used against ethnic minorities to intelligent AR glasses for police officers that can retrieve information about anyone the wearer looks at. This is not to say that we in Europe or the USA use Artificial Intelligence for purely ethical purposes. For example, the US military's Project Maven used artificially intelligent algorithms to identify targets of insurgents in Iraq and Syria. If a country or region takes the lead in AI, it will certainly derive some significant benefits from it. While currently China leads in investments and Europe and the USA lead in AI development, all face enormous economic inequalities that could negatively impact technological acceptance. The attitude
