AI Superpowers by Kai-Fu Lee

Academic AI breakthroughs hit headlines every month. Just in December 2020, a model surpassed the human benchmark for SuperGLUE (a performance score for several language comprehension tasks researchers are tackling today). Or earlier this year, an article published by the Guardian titled A robot wrote this entire article. Are you scared yet, human? went viral across Twitter and tech circles.

As someone wanting to enter the field of AI and computing, I had to ask myself: are these breakthroughs over-hyped? What does all of this research actually mean? And most importantly, how will it affect the future (and my potential job prospects)?

I turned to Kai-Fu Lee’s book, AI Superpowers to get some answers. Dr. Lee is an AI researcher (PhD from Carnegie Mellon University) turned industry expert turned venture capitalist. Throughout his career, he’s worked at Silicon Valley tech giants, like Apple and Google, and later started his own venture capital fund in China with an eye towards investing in Chinese AI start-ups. If anyone would have something interesting to say about the future of AI, it would be him.

AI Superpowers brought me these insights:

  • The US has had a lead in AI and information technologies, but whether that lead will be maintained or not depends on the nature of engineering/research roadblocks faced in the near future
  • The battle between the US and China has been described as an “AI arms race,” which may not be a helpful perspective to hold. There are many areas where collaboration and cooperation would accelerate the development of AI rather than competition (e.g. combining data sets)
  • Regardless of who comes out dominant in the space of ML/AI, there will be a further divide between the global north and the global south as China and the US capture much of the benefit from new AI/ML innovations, while countries in South America and Africa scramble to develop their nations
  • Performance for recent AI solutions are (generally) data driven, explaining US’s model surveillance capitalism and China’s various surveillance policies

What does all this mean?

After finishing AI Superpowers I’m left with some mixed emotions. Reassurance that my career will be stable and I likely won’t have to struggle to hard to find career opporutunities in the near to mid-term future. Some fear and anxiety at the potential political and social unrest caused by the implementation of various AI solutions. And a hint of optimism that we can get the engineering and ethical problems right – that we can bring about some kind of utopian vision of the future.

At heart, I like to be optimistic about things to come (even if current evidence might be a bit contradictory). But in the case of AI/ML, I honestly do think there are a large and growing amount of scientists, engineers, and politicians who recognize it and are giving it the right amount of attention.

I’d recommend giving the book a skim, especially with an eye towards the first 3-4 chapters outlining Kai-Fu Lee’s perspective on how China’s tech sector has grown over the years. As an American, it’s difficult to find sources that seem level-headed in evaluating China’s development and progress over the past decades. And although the bias isn’t absent from Dr. Lee’s work, it’s much better than works I’ve come across in the past. That alone might be enough reason to pick up this book.