My AI Investment Strategy

I barely survived the last AI Winter and intend to do better this time

Disclaimer: I am not a trained financial analyst. Investing involves the risk of loss. Do your own research before buying any stock. If you do not know how to research a stock and/or do not plan to monitor your investments closely, I suggest you invest in a diversified Exchange Traded Fund (ETF) such as SPY (SPDR), IVV (iShares), or VOO (Vanguard).

My Dad told me (several times), “learn from my mistakes so you don’t need to repeat them.” History, personal experience, a validated investment strategy, and a lifelong passion for understanding all forms of intelligence inform my AI investment strategy. Hear me out then make your own decisions.

History

"Gold rush" is a useful metaphor for describing the current AI frenzy. However, we can also draw lessons from the California gold rush itself. The discovery of gold in California in 1848 attracted over 300,000 miners from the United States and around the world. From 1847 to 1852, San Francisco's population increased from 500 to 150,000! As you might imagine, it took a lot of businesses to serve the needs of miners. The real winners—and persistent survivors — of the gold rush were the businesses that served miners. The German immigrant Levi Strauss provided miners with durable canvas pants. Henry Wells and William Fargo recognized the miner's need for financial services and founded Wells Fargo & Company.

Lesson for AI: Bet on suppliers and enablers, not application developers. Nvidia (NVDA) can hardly make enough AI chips and Super Micro Computer Inc. (SMCI) can hardly make enough data center rack computers to meet demand. However, both companies have high price-to-earnings (PE) ratios — too rich for my taste.

As for the AI 'miners', a good example is C3AI Incorporated (stock ticker AI). C3AI is an AI pure play. It doesn't have a high PE ratio because it has no earnings; it has been losing money for its entire three-year existence.

Half of the “49ers“ made a modest profit…in the beginning. But by 1855, the only profitable extraction was by medium or large enterprises comprising large workforces. I envision AI trending the same. AI applications are like video game development in that most people do not realize how much work is required to create software applications. Google (GOOG), Microsoft (MSFT), and Amazon (AMZN) already enjoy a scale and competitive barrier to entry that gives them an advantage over smaller AI companies.

I do own Google, Microsoft, and Amazon because they have built companies with untapped markets and large barriers to entry. Google has the dominant search engine (and AdWords to generate revenue), Microsoft has the dominant operating system, and Amazon has a dominant retail infrastructure. Each company has made significant investments in AI, which will only enhance their competitiveness, but none of them rely on AI to generate significant revenue.

Personal Experience

There is a good reason Silicon Valley is in California. The California Gold Rush drew optimists and dreamers. Some of their descendants found gold in technology and created Silicon Valley. I sold my second AI start-up company in 2000 during the last AI boom. I moved employees and family from Atlanta to Sunnyvale. Silicon Valley was in full "reality distortion zone" mode. When we arrived, you could not find a place to live. When I left less than two years later, you could not find a U-Haul truck.

This AI boom will bust too (see below). Cracks in the wall of AI hype get longer and deeper every day. When the bust arrives, there will be bodies everywhere. I do not intend to be one of them.

Investment Strategy

Buy low, sell high. A stock is expensive (it has a high PE) only if many people know about it, want to own it, and will bid the price up. Following stock tips from media sources means you are following the crowd. You are already too late. You will buy high and inevitably have to sell low. Bargains are found where crowds do not tread.

I cannot predict the peaks or the troughs of stocks or economies. I doubt if you can either. So don't try. I either seek quality stocks out of favor (cheap) or quality stocks with a long-term growth potential (like a MSFT or AMZN). A quality stock has a track record of profitability and financial stability.

I avoid companies that provide nice-to-have products and services. When the economy goes to shit, I want to be owning companies that deliver must-have products and services: food, energy, healthcare…did I leave anything out? So far, AI is a nice-to-have. But customer support is still nicer when humans do it. So maybe it is better to say that AI is only a potential nice-to-have.

Finally, I know some stocks I own may become worthless. That is why I never allocate over 5% of total investment funds to any single stock or investment. And since stocks within an industry or sector rise and fall together, I diversify across industries. Investing is hard work. It requires research and rational decision-making. Failing to assess risk and reward objectively is equivalent to betting. You will have better odds at a Blackjack table than betting your life savings on an unprofitable AI company (a so-called “story stock“).

I am a boring, long-term investor. Investing is most rewarding when divested of emotion. Otherwise, it becomes betting, a costly form of entertainment. I get my kicks from compounding. If you want to learn more about investing, visit the website of the American Association of Individual Investors at AAII.ORG.

Artificial General Intelligence verses Natural Intelligence

Don't get me wrong: I want computers to be smarter than they are today and I believe they will be smarter than us someday. Who wants stupid robots? Humanity provides enough stupidity for one world.

AI is to biological intelligence as internal combustion engines are to electric motors. Superficially, gas engines and electric motors do the same thing: they produce torque. Internally, they are so different that one cannot evolve or scale one into the other. Likewise, the mechanism of AI is incompatible with natural intelligence. Current AI technology cannot scale up to human-level performance. It is not possible. Achieving human-level performance requires an architecture and a process that is incongruent with current AI technology. Yes, it may be connectionist in nature, but the similarities stop there. BTW, this subject is the dominant theme of my upcoming book.

Success has painted the AI industry into a corner. Demonstrations of self-driving cars, the beating of world-class chess and Go players, and conversational chatbots are impressive. Consequently, substantial funds are being invested in AI to achieve human-level performance or Artificial General Intelligence (AGI). But this is a fool’s errand. AI does not scale into natural intelligence any more than internal combustion engines can scale into electric motors. And like the transition from gas to electric automobiles, inertial resistance, existing infrastructure, vested interests (investments in AI companies), and the cost of employee retraining works against transitioning out of the current popular paradigm. As a result, we will continue to invest in AI—once again—until it reveals its rotten core. I do not know when that will occur, but am confident that it will. There will be another AI winter.

The good news is that another AI winter might motivate computer scientists to question their approach to intelligence. They might even pick up a biology book. Maybe my book. But there will be another boom cycle. They will use silicon to implement natural intelligence. It won't be a general algorithm because evolution results in kluges, not elegant algorithms. It will be artificial at the level of implementation only: microelectronic circuits instead of nerves. But robots will duplicate the algorithms, computational frameworks, and environmental embodiments found in nature. So please don't call it AGI. But get ready because it may be the last wild ride of its kind.