AI Superpowers

Last updated Aug 28, 2020 Edit Source

• Book

I got my third recommendation for AI Superpowers a couple of weeks back (#iykyk), and conveniently had many copies lying around the house, so I decided to pick up this book. I ended up zipping through this book; I feel like it only took me around four hours to read?

There were a lot of things that I loved about this book, including the comparison of the Chinese and American startup ecosystems. Kai-Fu Lee also did a really great job of unpacking the things I’ve observed anecdotally into concrete theory. I had realized the street pedlers in China ask for money via Wechat Wallet, but I did not connect that intuitively with China’s leap in mobile payment platforms.

These were my three main ideas:

The industrial revolution fueled by electricity is similar to today’s AI revolution, which thrives on an ecosystem of data, entrepreneurs, AI scientists, and AI friendly policy. What will be the new Standard Oil, and who will be the new John D.?

Why?

• The US government deliberately takes a laissez-faire approach to entrepreneurship and is slashing funding for basic research. The Chinese government invests proactively and lends funding + legitimacy to entrepreneurship.

• Profit hungry entrepreneurs mimicking competitors ➡️ relentless iteration, cost control, flawless execution, raising funds at exaggerated valuations

• Different goals in entrepreneurship: Silicon Valley: liberal arts + innovative thinking ➡️ clean digital platforms; China’s scarcity mentality: getting rich ➡️ implementation heavy approach, vertically integrated, recruiting users

• American startups favor the “light approach” - sharing information, closing knowledge gaps, connecting people digitally; Chinese startups favor the “heavy approach” - recruit each seller, handle goods, run the delivery team, control payment (ex. mobile payments market)

If AI will replace 40-50% of jobs in America within the next 15 years, we need government leaders who are well-versed enough in technology to respond to the impacts of this change. In developing countries, AI-driven automation will undercut their historical economic advantage: cheap labor. Smart cities like Hebei’s Xiong’an will shift power from government to tech companies.

In the end, I had two big takeaways from AI Superpowers.

One, which Kai-Fu Lee also spoke about in his TED Talk, is the idea that AI, rather than merely eliminating jobs, will give us the freedom to do the work that makes us human. AI can optimize but not innovate. The jobs that will be most protected are creatives and caregivers.

Two, to always put family first, and to never let my work ethic take precedence over love for my family. I hope I will not need a terminal illness to remind me to prioritize eulogy virtues over resume virtues. No matter how influential I become, the people closest to me are the ones I will have the most impact on

Overall, this book was a fascinating read. My goal in the future is to be well-versed enough in technology to be dangerous. Maybe that will be investing, maybe that will be taking companies public. In either case, I want to use my strengths as a Third Culture Kid. We’ll see. :)

• Alibaba: The House that Jack Ma Built, Duncan Clark

• Folding Beijing (北京折叠), Hao Jingfang (郝景芳)

• I really want my Chinese to be good enough so that I can go back to reading in Chinese again :/

People:

• Marvin Minsky

• John McCarthy

• Herb Simon

• Geoff Hinton - deep learning, computer vision

• 李克强 (Li Keqiang) - 大众创业, 万众创新(双创) - Mass Entrepreneurship, Mass Innovation

• Connie Chan, Andreesen Horowitz

• Angela Wang’s  TED Talk on Chinese e-commerce from October 2017

• Sebastian Thrun - Kitty Hawk, Udacity, Google X (still has strong ties to the university for sure)

Things:

• Hidden Markov models

• deep learning, reinforcement learning, transfer learning

• Lean Startup Model (27) - rapid product development cycles to see if a proposed business model is viable; business-hypothesis-driven experimentation, iterative product releases, and validated learning

• mission-driven (new idea or idealistic goal with clean mission statements) vs. market-driven (make money, adopt any model, go into any business)

• Moravec’s Paradox (166) - high-motor activities are the most difficult to automate

• O2O - “online-to-offline” - turning online actions into offline resources (Meituan - ride-hailing, food-delivery)

• The four waves of AI:

  • internet (data) - AI algorithms as recommendation engines run on labeled data

  • business (finance/banking) - trading algorithms

  • perception (face/voice recognition) - blended environments (Online-Merge-Offline), the next step after O2O

  • autonomous (robots) - ex. self-driving cars, packing robots

• Who took Alibaba public? Daniel Fertig (Columbia Law ‘02) of Simpson Thacher & Bartlett, Jay Clayton of Sullivan & Cromwell (Now Chairman of the SEC). What was the process like?

• Where is China’s Sand Hill Road? I don’t think it’s 中关村大街