Why the AI Race Depends on Math

July 27, 2020 - 7 minutes read

China has taken artificial intelligence (AI) and ran with it. The country started using the technology for facial recognition and law enforcement in 2017. Since then, what used to take AI minutes to solve now takes only a few seconds.

But how did China’s AI become so fast so quickly? With constant tweaking of AI algorithms to improve their output and continuously adding in new findings from the math field, an algorithm can become supercharged overnight. With these developments, the USA only has a short amount of time to catch up or it risks being forever left behind China in the AI race.

The Right Foundation for Success

Whichever country has the best AI technology is set to become a stronger world power than ever before. The country in question would have the strength to shape global commerce, finance, war technology, computing, and telecommunications.

And since a strong mathematics foundation equates to better AI, there’s a very real possibility that the country with the best education in this subject could come out on top. America has long lagged in STEM education. In fact, it has gotten so bad that pouring millions of dollars back into our perpetually-defunded public school systems won’t fix the problem anymore.

Strong math teachers are needed. But many current educators in this field missed learning an adequate foundation themselves while they were a child. The cycle will continue to repeat itself unless we overhaul the entire nation’s math curriculum, which is much easier said than done.

Because AI works off of advanced math and statistics concepts, we’re experiencing a major gap in our knowledge. It turns out that foundational concepts that AI is built on are skipped over in calculus class. And it’s not just AI that’s going to suffer: machine learning applications also require a strong foundation in math and computer science.

The Proof Is in the Math

American secondary and university students are falling behind their peers across the globe. 15-year-old students came in 35th on the OECD’s 2018 Program for International Student Assessment. That standing is below the global OECD student average. College students are also facing trouble due to professors prioritizing memorization strategy over true learning.

Computing platforms, from iPhones to the world’s most powerful supercomputer in Tennessee, use complex math and electrical engineering concepts to find answers rapidly. Some AI experts argue that the country with the most amount of data will win the AI race, but many others disagree, saying that data without math knowledge is useless. They say that, unless you know how to use the data in math applications and algorithms, data will only take you so far.

New math findings are focused on how to work with datasets that are sparse or contain partial information. Programming an AI algorithm to throw out useless information is complex and difficult, a journey that goes through multiple branches of math that compound on top of each other.

It’s not possible for even STEM graduates to jump straight into AI because it takes years of math learning to get to a level high enough to know what’s really going on in an AI application. It’s not just the hard math that students are lacking: they’re missing the core problem-solving skills that would be required to excel in a math program.

A Cascading Effect

There’s another major issue, and it’s one that the USA is responsible for perpetuating. US colleges’ Masters and Ph.D. programs are largely international student-based. Because US children are lacking confidence and knowledge in their math foundations, fewer and fewer apply for graduate education in math or a math-related field.

In 2017, according to the National Science Foundation, nearly 70% of Master’s students and over 64% of Ph.D. students in US computer science programs were foreign. That year, half of all mathematics Ph.D. degrees were given to international students. Most of these students are Chinese and Indian. The education systems in these countries are lacking, but many of these students will return back to their country to accelerate research and development in emerging technologies like AI, IoT, 5G, and more.

Another issue caused by our own government is the stringent visa requirements for skilled workers. Many students are forced to return to their home country upon graduating because it’s difficult for them to obtain a visa even though their skills are in high demand.

At the moment, China is investing heavily in AI technology and the foundations for the technology (like math, computer science, and electrical engineering) to reach its 2030 goal of becoming the global AI leader. It’s also pouring money into other technologies in an effort to meet its 2025 goal of becoming the dominant power in most emerging technologies.

AI’s Future Is Foggy

China’s federal government recently pledged $2.1 billion to develop an AI industrial park outside of Beijing. Huawei, a Chinese mobile communications company based in London and Singapore, is at the forefront of this expansion with new releases of next-generation “AI processing” chips.

It’s all paying off: China’s AI market is worth around $3.5 billion today, and Beijing wants to reach an AI market worth $142 billion by 2030.

Because of China’s clear goals and plans, the country has a confident and clear path to AI domination. But as we’ve seen with their past uses of AI in policing their citizens and creating a “social credit” system, the US would be remiss to not take charge of AI and use it for social good.

Tags: , , , , , , , , , , , , , , ,