China marching forward with artificial intelligence
I have visited nuclear power plants, seen artificially grown ears being implanted into goats, had my fair share of meetings about semiconductors, carbon fibers, gene therapy, solar energy and even a lecture about the science behind the chewiness of noodles (yes, that’s a thing).
But if you ask me what’s the most radical technology, the one that will effect economies all over the world:
Artificial intelligence (AI) is the single most game-changing technology out there.
If you don’t believe me, then please read on. If I haven’t convinced you by the end of this article, then contact me and let’s have coffee. I will be curious to hear your counter-arguments. For those who already agree with me: read on as well and find out why China recognises the importance of AI.
Humans have always empowered themselves with new tools. Some were physical, such as spears, hammers, and steam engines, while other tools were cognitive concepts, such as language, reasoning and mathematics.
It is easy to forget how important access to tools or technology really is. Imagine that companies in your country did not have access to the latest computers. In today’s globalised world, it would be a disaster for their competitive position. Yet, this is the situation China has been in for decades. Companies in Western countries had access to state-of-the-art equipment. They could buy it, use it or improve it, but China was often banned from acquiring the latest technology. This made them very aware of the effects of lagging in technological development.
Now consider that the next wave of innovation does not come from hardware (microprocessors), but from software (algorithms and data). What if your country does not have access to the latest software innovations?
Most of us are stuck on thinking about ‘humans versus artificial intelligence’ (and that is a valid discussion!). But long before we invent AI on par with our own capacity, AI will already change economies all around the world. The reason?
Humans and companies using artificial intelligence will outperform their competition.
Artificial intelligence has had its ups and downs. In the 50’s everyone got excited. But in the 70’s many people lost their interest. Then when ‘machine learning’ got (re)invented, and AI as a whole got rebranded as such, the world started paying attention again. Instead of pre-programming every decision model, the new AI systems could learn to create their own decision models. More recently, a subset of machine learning called ‘deep learning’ has created some very cool breakthroughs and drove up investments in AI even further.
Yet, still few of us really understand how important AI is. Think again about what your world would look like without computers. A logistic company without navigation systems, a bank keeping account records only on paper, a hospital where you can’t get an CT or MRI scan?
Major mobile operating systems like Windows, Android and iOS all come from one country: the United States of America. What if something similar happens with AI, but then the countries or companies developing those AI systems decide not to share their inventions?
Computers are an enabling technology. Just like windmills, ships, steam engines and
electricity were before. In the 17th century the Dutch were the best shipbuilders. Combined with access to cheap energy from wind mills, this enabled them to become the biggest economic, scientific and military power in the world. One ship with crew outperforms thousands of men, if you want to trade with Asian countries. You can’t just carry those goods back and forward. Productivity didn’t jump with a few percent, it went up by factors. That is why it is so important to understand that:
Enabling technologies exponentially increase the output of humans.
Now imagine that AI or machine learning is the ship technology of the 21st century. Whatever challenge you face or opportunity you see, you can empower yourself and your staff with machine learning. No matter how many competitors you have, as long as you have better machine learning you will outperform them. In virtually any task, since machine learning is a much more versatile enabling technology than ships were.
So what about all the other technologies being developed?
- 3D printing? Who’s going to help you design those complex designs? Machine learning is.
- Genetic research? Who’s going to help you make sense of all the genetic data? Machine learning is.
- Telecommunication? Who’s going to help you create the best new antenna and chip designs? Machine learning is.
- Medical technology? Who’s going to help you identify cancer better and faster? Machine learning is.
- Electric vehicles? This industry will basically completely run on machine learning. Machine learning will make driverless cars possible, making transportation much more energy efficient and safer too.
Should I go on?
How does China fit into this story?
Well, as I explained before, China has had the experience of not having access to new technologies. This makes them a lot more keen to figure out what is going to be the next key enabling technology. It should therefore not come as a surprise that:
China is conducting more scientific research on ‘deep learning’ than any other country in the world.
And in case you are wondering, “But what about quality?”, know that China also has the most cited scientific papers on deep learning. Meaning that the most referred-to scientific papers are Chinese ones, an important indicator for quality.
China continues to support more academic research on deep learning, not just in absolute numbers, also as share of the total amount of funded academic research projects.
In the chart below we can see that since 2010 there has been a sharp increase in funded research projects related to deep learning in China. This increase is even outpacing the overal growth in academic funding in China significantly, as the share of ‘deep learning’ research projects is growing over time.
In patents there is a similar trend taking place. Chinese patent applications focused on artificial intelligence are growing fast, at rates outperforming all other nations.
In recent years we see Chinese invention patent applications on ‘deep learning’ grow with 300% per year.
Is this a call to action?
For governments definitely. Many sectors will be affected and AI will change competitive landscapes much faster than before. Internet cables don’t care much for borders, AI-fuelled industries won’t either. (Think about the global market shares of Google, Facebook, etc.) To prevent its industries getting into serious problems, governments need to double down, triple down or even quadruple down on AI, now. Invest in high-risk-high-reward research. Encourage science-industry cooperation, and make sure there’s an ecosystem for startups in AI (and big data). Invest in future generations, by incorporating concepts of software development, probability and machine learning at early ages in your educational programs. Yes, some of you might think “Jaap is insane”, but there are Chinese primary schools where they already teach children these basic concepts (there are also similar initiatives in the US and the Netherlands).
Companies should keep track on the progress of AI and specifically machine learning in their sector. Machine learning has been moving out of the academic world and into real-life applications. Data scientists are paving the way for data engineers, applying machine learning algorithms in any sector. Work with technology foresight specialists to devise scenarios about how your industry would look like ten years from now, when machine learning will be embedded into it.
On a personal level you have much more freedom. Machine learning is a technological trend, not a cult. There are many other things to do and to explore. AI still has to have something to enable, it doesn’t do much on its own (except when we create artificial general intelligence, but that’s a whole different conversation). However, just as putting “MS Word” on your resume is not something I would recommend nowadays, so will “Experience with Cortana/Siri/Echo” become a similarly obvious skillset in the future.
Jaap van Etten – May, 2017
Article originally posted (in Dutch) on www.china2025.nl
Photo: May 11, 1997. Twenty years ago. Deep Blue versus Garry Kasparov, Game 6. That day Deep Blue won, and with a score of 3.5 versus 2.5 became the first computer to beat the human world champion. … Today you can buy a chess engine for your laptop that would beat Deep Blue quite easily.