Value Creation for A.I. in the Era of Implementation
By Michael J. Donovan
Changsha, China (November 21, 2018) - The evolution of artificial intelligence (“A.I.”) and cognitive computing systems has been anything but smooth. Mathematical and theoretical applications for A.I. have been worked on vigorously for the past few decades. However, only in the last five years have real-world applications begun to come to fruition. Proofs and models in the 80s and 90s were able to be drawn out on paper, but when datasets were on the scale of megabytes, it was difficult to realize the true value of these models. We are many ways in the midst of a renaissance of neural networks. Three critical pieces of the A.I. puzzle are being put into place: economies of scale, computing power, and availability of data.
The United States has been a dominant leader in A.I. since its inception in 1955 when John McCarthy coined the term. American A.I. leaders have paved the road for us to enter the Fourth Industrial Revolution, where the implementation of knowledge will be far more valuable than its inception. Since we in the U.S. have always been dominant leaders in computer science, it is difficult for many of us to fathom that a newcomer like China could possibly be an A.I. superpower within a decade. In the age of A.I., those with massive amounts of data and the ability to efficiently organize at the national and global levels have a number of advantages. It’s said that if data is the new oil, then China is the Saudi Arabia of data. How the United States and China collaborate in the field of A.I. will be fundamental to its global development and profitability for humankind.
We are entering an era when A.I. is being more application-driven than research-driven. This is when data becomes king. Machine learning and deep learning are currently two of the most important branches of artificial intelligence and are experiencing rapid growth. They allow computers to learn from data and experience. Although many of the critical breakthroughs in machine and deep learning have been discovered in the U.S., China is the arena where these innovations are rapidly being implemented. China may not have the lead when it comes to super experts; however, when it comes to general experts, China has a very large talent pool. General experts are more than capable of utilizing machine learning algorithms effectively in order to develop very capable systems. China has already become a leader in image and speech recognition.
For ambitious companies in A.I., China is no longer just an option. It represents an ecosystem where algorithms can be optimized at scales that are impossible to experience in the U.S. A mobile first market with 1.4 billion people has volume, velocity, and variety that goes a long way. This is quickly being evidenced in the field of healthcare. China is looking to A.I. as a solution to many of its healthcare ailments. China’s hospital network is currently coming together to share data in ways that will accelerate the implementation of A.I. in its healthcare system. This coupled with the seeming inability for us in the U.S. to overcome the fragmentation of our healthcare system means that China will be the place where new A.I. applications for healthcare will be initially implemented. Patient diagnosis, public health analysis, surgery, precision medicine, drug discovery, and insurance coverage all have the potential to be dramatically reshaped.
Implementation of A.I. in Healthcare
Technology is empowering patients and doctors more than ever before. Eric Topol’s book The Patient Will See You Now discusses how technology is redefining the doctor-patient relationship. However, technology alone cannot bring about a healthcare revolution. A concerted effort needs to be made in order to have patients’ data shared securely and effectively. Without this data, A.I. algorithms will not be able to perform with superhuman capability. China recognizes that these algorithms need to be rolled out cautiously. It also recognizes that it does not have the luxury of spending years debating how this data should be utilized.
China has not been shy about making A.I. a national priority. In July 2016, China issued a notice regarding the application and development of big data in the health and medical sectors. One hundred (100) regional clinical medicine data demonstration centers will be built across the country.  These hospitals will generate an incredible amount of data. China’s hospital staff are particularly burdened by the task of treating China’s large population. No group of doctors is cheering for the success of A.I. in the medical field more than the Chinese. It is common for one doctor to see 80 - 100 patients per day. The incredibly overburdened Chinese healthcare system is rapidly looking to the field of A.I. for applications. Radiologists are beginning to be aided with A.I. programs. Computers are being trained how to scan CT and MRIs in order to spot abnormalities. Recently, as published in Nature Biomedical Engineering, a group of Chinese researchers fed a computer thousands of images from colonoscopies in order to have the computer teach itself how to recognize polyps that may indicate the presence of colon cancer.  Datasets like this are quickly being regularly analyzed.
One beautiful aspect of powerful machine learning algorithms is that objectives are given to an A.I. system and then the system teaches itself how to perform a task with superhuman capability. For cancer imaging, doctors can feed the algorithms images where thousands of patients’ tumors are indicated and the algorithm can subsequently teach itself how to identify such targets. The algorithms are optimized when more data is processed. In the OMICs fields (genomics, proteomics, and metabolomics), analyzing the correlations between massive datasets will need to be done via A.I. Getting one’s genome sequenced has dramatically fallen in price. However, post analysis of one’s data is still quite expensive. A.I. will help make post analysis of one’s data become affordable for the average human. An ecosystem that can provide volumes of datasets with millions of different datapoints is ideal for machine and deep learning. Healthcare has these characteristics. However, silos of data that cannot be analyzed and shared securely will hinder the growth of A.I.’s capabilities.
Some doctors in the U.S. are experiencing a certain degree of hesitation when it comes to implementing A.I. A great deal of wealth will be created by A.I.; however, there will need to be restructuring in our job outlook. The high-valued jobs will be composed of people who perform jobs of compassion and creativity while utilizing A.I. at their core. The best doctors of the near future will be the ones who actively utilize A.I.
Moreover, it could soon be considered negligent for a doctor to not utilize an A.I. program when calculating the appropriate drug cocktails and regiments for patients. Teams such as Dean Ho’s at the National University of Singapore are utilizing A.I. programs to determine the appropriate drug dosages for cancer patients.  The generated plans for the patients have already been leading to impressive results. When these algorithms are generating better results than human judgement alone, one must begin to insist that the algorithms are more widely utilized and tested. Teams with similar objectives are testing models for predictive A.I. For drug discovery and development, algorithms based upon generative adversarial networks (GANs) are being utilized as key tools. Furthermore, conditional GANs are being developed for the analysis of gene expression interference. A.I. conjured, novel synthesis protocols are leading to new materials that can interface with biology in novel ways. The PhD student in a decade will require a certain level of understanding of biology or chemistry and computer science or A.I. engineering.
A.I. will ultimately lead to value created for industries that do not exist yet. We have the potential to make the world better if A.I. provides services for the greater good of our society.
Redefining Innovation and Embracing New Business Models
One of the critical differences between American and Chinese technological ecosystems is the intense competition and speed of iterative progression. Many tech leaders who work within the Silicon Valley ecosystem believe that true innovation only comes from bold, original ideas. However, in the Chinese ecosystem, iterative progression has often resulted in innovation. Companies that began as a copycat of a U.S. version have generally morphed into applications that are very unique. Within 7 years China has gone from an emulator to an innovator. Applications that are developed in the U.S. and then allowed to evolve in the Chinese ecosystem, often result in superior products. This is evidenced by WeChat, MeiTuan and Tmall. When it comes to the world of A.I., this is extremely advantageous because A.I thrives on iterative progression.
China’s ecosystem is quickly evolving to be very unique and connected. In many areas, it is highly original. Startups are now taking root from bold and original ideas. Therefore, there are unique products being developed solely for the Chinese market. In A.I. alone, there are approximately 15 A.I. focused Chinese unicorns. The utilization of big data, development of core infrastructure, and harnessing of A.I. are fundamental plans in China’s national strategy. When similar initial algorithms are utilized, a company that develops outside China’s ecosystem, without utilizing the same breadth and depth of data as one that has developed from within, will have a difficult time performing as accurately. Being able to operate at the economies of scale within China will allow for entirely new areas of industries to be forged.
When it comes to the world of the internet, we are already beginning to notice two very distinct ecosystems emerging. There’s an internet of data being led by the United States and then there is an internet of data being led by China. Peter Thiel’s book Zero to One describes a playbook that results in wins in the Silicon Valley ecosystem; however, in China, this playbook would most likely result in little success. With the scale of the Chinese ecosystem and tenacity of Chinese entrepreneurs, companies must embrace competition. Every new idea that emerges in China is quickly met with intense, iterative competition. The global A.I. behemoths of the future will be Zero to One minded companies that allow themselves to go through iterative progression at a scale that currently only the Chinese ecosystem can provide. This brings us to the critical role infrastructure plays.
Infrastructure for A.I.
The speed and efficiency of the rollout of the required infrastructure for A.I. are matters of both city and national interest. 5G networks are critical to the success of secure and fast transfer of pertinent data. With a national 5G network being laid out in China, companies such as Intuitive Surgical (HQ in California) have many opportunities in China that are simply not in the U.S. at the present stage. Starting next year, 5G-connected pilot cities will be the laboratory for A.I. and robotic teams to stress test their systems. Medical robots fueled with A.I. capabilities and communicating over 5G networks represent exciting developments in healthcare. Remote surgery with both A.I. and minimal human guidance can finally become a reality.
The amount of investment being put into 5G infrastructure across China in order to have a modern internet of things infrastructure is reaching a level of development that many analysts are calling China’s “5G tsunami.” According to Deloitte, “China has built 350,000 new cell sites, while the U.S. has built fewer than 30,000 since 2015.” From 2020 to 2030, mainland China is estimated to spend US$411 billion on 5G mobile networks. According to Deloitte, it is estimated that the equipment necessary to add a carrier in China costs about 35 percent less than the U.S., suggesting that Washington would need to spend 2.67 times the amount that China does to generate an equivalent amount of wireless network capacity.  This is important since in 2035, 5G is expected to enable US$12.3 trillion of global economic output.  As we enter the Fourth Industrial Revolution, the countries, companies, and cities that are able to capitalize on A.I., the internet of things, and advanced mobile networks (e.g. 5G) will lead in generated revenue.
A.I represents the most revolutionary technology in modern human history. Yet, in order for it to advance, it needs a collective effort across large populations. The concept of value creation is changing considerably in an open-sourced world. Information now needs action in order to lead to value.
Major breakthroughs are still required to progress beyond the current, elementary stages of A.I. We’re creating superhuman applications but are currently limited in multidomain functionality. The U.S. is leading in basic research; however, China is creating a great deal of wealth in the implementation of A.I. The U.S. will need to rethink its speed of development in order to ensure its companies harness a significant portion of the value created. We in the U.S. are no longer just competing with teams from Europe and Japan. Chinese speed and scale of implementation are on an entirely different level.
The U.S. will likely lead the developed world in A.I. while China leads the so-called developing world. Yet, we in the U.S. must embrace competition at the level that China operates. This is the new normal. American teams must be just as agile as Chinese ones. This requires collaborating and working with them. Action and organization in the U.S. need to occur expeditiously.
The U.S. can very likely extend its lead in the academic realm of A.I. with new models and algorithms. Journal articles and accolades will be given, but value created for the university and local community is not guaranteed without proper implementation. The speed at which we in the U.S. repair public infrastructure, implement electric charging stations, and build high-speed rail can not be the speed at which we implement A.I. if we are to be leaders beyond 2030.
1. The State Council. The People’s Republic of China. China to boost big data application in health and medical sectors. 2016 http://english.gov.cn/policies/latest_releases/2016/06/24/content_281475379018156.htm.
2. Wang, P., Xiao, X., Glissen Brown, J., Berzin, T., Tu, M., Xiong, F., Hu, X., Liu, P., Song, Y., Zhang, D., Yang, X., Li, L., He, J., Yi, X., Liu, J. and Liu, X. (2018). Development and validation of a deep-learning algorithm for the detection of polyps during colonoscopy. Nature Biomedical Engineering. Available at: https://www.nature.com/articles/s41551-018-0301-3.