When I first read Kee’s Chinese books last year, I was moved by his gentleness and humility. However, his English writing, “AI Superpowers” revealed another side of him: poignant and piercing. He argues confidently in favour of the Chinese AI progress. His analysis boasts of technical soundness and visionary zeal.
His central argument is that we are in the age of AI implementation, and not discovery; China, with its sheer amount of data, would eventually surpass the USA in terms of AI progress. Although we are constantly pushing the capabilities of AI, we are still decades away from the next major breakthrough. Currently, only “narrow AI” works well — AI only outperforms humans in extremely restricted settings, and is still far from “general intelligence”. The breakthrough of Deep Learning came in 2012 with Alex Net, after decades of AI winter. Most likely, the next breakthrough, including but not limited to discovering the consciousness of AI, imbuing it with “common sense” and natural language processing capabilities, would still be decades away. Since we are in the age of implementation, China, with an army of great (but not elite) AI engineers, can be more productive than Silicon Valley, which has a small pool of AI elite AI engineers.
One key drawback of this book is Lee’s proposal for future economic policies. He believed that to shoulder the responsibility of job displacement from AI, we need to introduce a “universal basic income”. However, I see his analysis of lacking concert backings as he is neither trained in economics and politics. His proposal lacks solid data and predictions, making it rather weak, and has ample room for improvement.
Xiaohao's ruminations 