Over the past year, you’ve likely seen and heard (or even used) something involving artificial intelligence (AI). There has been wide speculation about how this technology might affect humanity and many important questions have been raised. Is AI a hype cycle? Is it the next industrial revolution? Or is it science fiction come to life to upend life as we know it?
This is the first in a series of blog posts looking closer at AI, in particular what it means for your portfolio, and what it could mean for our overall future. If you’re unfamiliar with AI or new to what it can offer, I’ll also be introducing you to some of my favorite AI tools along the way.
American Productivity in Decline
First, a brief anecdote. I was riding horses on a high mountain trail in Montana with a Treasury Department Deputy Assistant Secretary. He wasn’t the Fed Chairman or anything like that, but this Harvard educated economist was the best vacation friend an economics nerd like me could possibly have met.
It was one of those small world moments, too. We pieced together how he was once good friends with my cousin from Short Hills. Coincidences aside, we talked about the national debt, the coming explosion of entitlement expenses (e.g. Medicare), and a puzzling lull in productivity growth.
Cowboy hats. Big sky. Economics. I was in heaven – except for our conclusion: America needed productivity to giddyup up, and fast. Eight years after that vacation, it still has not. And we absolutely need stronger productivity gains to avoid raising taxes or cutting entitlement benefits.
Want smart updates on the world of money?
Enter AI
We have actually seen artificial intelligence coming at us for a while. While short-term market moves are still (and likely always will be) a murky business, there are plenty of examples to support the notion that long-term trends can be observed in advance, if not their full significance. People saw climate change coming, right?
In 1896, Swedish scientist Svante Arrhenius predicted the greenhouse effect. Old Svante didn’t know the exact contours or implications of his observation. But he was able to predict that industrial emissions of carbon dioxide were significant enough to cause global warming. If Svante had developed an investment thesis to profit from that observation, he may well have fallen flat on his face. But the point is, some macro trends are observable even if the exact contours of their economic implications remain mysterious.
In 1965, Gordon Moore predicted the coming exponential growth in computing power in tandem with an equal but opposite collapse in prices. What came to be known as Moore’s Law – the doubling in silicon-based computing power paired with a halving of prices every two years – has held up, with only some slowing in the past decade or so. Point is, Moore also saw the trend.
Ray Kurzweil, another futurist, saw the same trends in computing and made a number of prescient predictions as early as 1990, including pocket-sized personal digital assistants, cloud computing, voice recognition, drones, universal translation, and streaming. Not bad.
In 1999, Ray predicted The Singularity. The Singularity refers to the moment artificial general intelligence (AGI) exceeds human intelligence, sparking rapidly accelerating technological progress beyond human control around 2045. Fun! Ray now says 2099. Pheew? Maybe he just extrapolated too far. But the idea of machine intelligence is quite old. Rene Descartes contemplated intelligent machines in 1637. Since then, the pace of AIs emergence (in our consciences at least) keeps accelerating:
- In 1950, Alan Turing devised a test to identify machine intelligence
- The Terminator first screened in 1984. I was 13 and impressionable
- In 2011, IBM’s Watson won a champions’ tournament on Jeopardy
- In January of 2023, ChatGPT-3.5 took, but failed, the bar exam
- In March of 2023, ChatGPT-4.0 passed the bar exam, scoring in the 90th percentile
The Risks of Present Day AI
Have machines passed Turing’s test? Can computers fool humans to think they are interacting with a human? While no one thinks we are on the cusp of AGI, the dangers of AI are being taken seriously by those sitting in the rooms where AI is happening.
In May, the Center for AI Safety released this starkly brief statement:
“Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.”
The statement was signed by 350 leading business and engineering minds working in artificial intelligence including Sam Altman, the CEO of OpenAI, the company responsible for the ChatGPT chatbot that got everybody talking in late 2022.
I do think something is coming.
Like nuclear power, it holds both promise and peril. It certainly has society’s attention. Google searches for ‘artificial intelligence’ have increased significantly since the fall of 2022. Even I, while writing this article, used AI to momentarily offset talk of societal-scale risks with a little whimsy. I prompted my favorite AIs to “Write a haiku about AI taking over human jobs.”
Haiku by Open AI’s ChatGPT-4.0
Jobs in silicon hands,
Human tasks in circuit’s dance,
Echoes of progress.
Haiku by Anthropic’s Claude.ai
Silicon brains grow fast
Human roles soon obsolete
Machines now in charge
Haiku by Google’s Bard
The machine is fast
It works without rest or pay.
Where will we all go?
Before you despair, note Claude’s extra syllable in the first line of the haiku. Well, you could pronounce it Sil-con. Despair!
Or consider the tale of the lawyer who found himself in hot water with a judge after relying on ChatGPT to research a legal brief. The cases he cited were entirely fabricated by AI. Convincing but nonexistent, the citations ChatGPT returned were formatted correctly and even ‘written’ in the style of actual judges. They were fiction.
Your Honor was not pleased.
Tips to Avoid AI Mistakes
The fault, in fairness, lies with the lawyer.
OpenAI discloses that, “ChatGPT may produce inaccurate information about people, places, or facts.” ChatGPT pointed that out to me when I asked it to critique this post for factual inaccuracies. Decent point! Lawyers are supposed to read the fine print.
Chatbots, it turns out, are prone to ‘hallucinate.’
The large language models on which they rely on are simply using probability to predict the next word. Large language models primarily make it ‘sound right’. I frequently find inaccuracies and won’t rely on AI answers without cross-referencing them. Don’t expect logic or reasoning. And confirm, confirm, confirm.
So, is AI overhyped or are we on the cusp of revolutionary change?
Yes and yes.
Amara’s Law states that in the short term we overestimate new technologies. In the long term, we underestimate them. Using the internet as a metaphor, we are using dial-up modems today.
ChatGPT thinks I am overstating our current lack of sophistication and insight. “The analogy ‘Using the internet as a metaphor, we are using dial-up modems today’ could be misleading. While it’s true that we are in the early stages of leveraging AI technologies, it might imply that our current understanding and usage of AI is slower or less sophisticated than it actually is.”
Typical AI.
It certainly seems intelligent. It makes you wonder what underlies our intelligence?
Regardless, the potential creative destruction of AI use could be on par with our country’s transition from agrarian to industrial. That would be very disruptive to some workers like lawyers (or even financial advisors). But it is hard to argue that we are materially worse off today than we were before tractors. It’s these types of tradeoffs that will likely define the financial future.
AI and Your Investment Portfolio
So, what does AI mean for your investment portfolio? (That’s probably why you’re still reading this.) Artificial intelligence still pales in comparison to the informational content of the market’s AI – aggregate intelligence. All information known by all investors is reflected in market prices through the process of buying and selling hundreds of billions of dollars worth of securities every day. The data-informed financial models driving those trading decisions are not available on the open internet and so not available to AI.
That is not to say that active investors have not employed AI processes for years in hopes of gleaning an informational edge. By scraping text from financial reports or gauging social media sentiment, they aim to identify stocks set to outperform. Any material insights they might note are quickly reflected in market prices. As more investors employ the same tools, any brief advantage diminishes.
Most damningly for those who might fantasize about outsmarting markets, AI’s forecasting prowess requires stable patterns. My phone’s navigation app recognizes my commute to work because it follows a routine. Autonomous vehicles halt for familiar stop signs. But stock markets change day to day in complex and dynamic ways. AI trying to predict market prices is like self-driving cars trying to read stop signs with words, shapes, and colors that differ from one day to the next.
AI will almost certainly continue to enhance investment processes and its impact will go well beyond the financial sector. McKinsey’s June 2023 report entitled The Economic Potential of Generative AI: The Next Productivity Frontier exclaims, “Our latest research estimates that generative AI could add the equivalent of $2.6 trillion to $4.4 trillion annually across the 63 use cases we analyzed – by comparison, the United Kingdom’s entire GDP in 2021 was $3.1 trillion.”
The tl/dr version: McKinsey predicts AI will be worth a G-7 country, annually.
AI and Your Financial Future
It does not take much intelligence, artificial or human, to imagine how AI could boost productivity, economic growth, and your portfolio. Or we may all be pets to the machines one day. Who knows? Like all good financial advisors, we will watch, learn, and plan accordingly.
ChatGPT critiqued my last paragraph:
“Lastly, the whimsical note about ‘we may all be pets to the machines one day’ is a popular trope in discussions about AI, but it’s worth stressing that this is a far-fetched scenario based on speculation, not on current AI capabilities or trends. Even with advances in AI, there are significant challenges and ethical considerations involved in developing machines with that level of autonomy and power.”
If you say so, ChatGPT.
