I’ve been thinking a lot about ChatGPT and the exploding world of large language models (LLMs), of which ChatGPT is the most well known. On the whole, I find LLMs to be an exciting new technology that will transform the way we interact with data and technology. But I do have my concerns.
If you’ve been following me for any length of time, you’ll not be surprised to learn that I’m not particularly impressed with the extravagant claims of the AI builders about how we’re all about to be made obsolete. I beg to differ.
It was amazing to listen to Lex Fridman interview ChatGPT founder Sam Altman. I love Lex Fridman and I admire Sam Altman, but it was embarrassing to watch their hushed-toned faux-humility bromance as they pondered which future version of ChatGPT was likely to achieve artificial general intelligence (AGI) and consciousness, and which pantheon of luminaries Sam Altman would be remembered amidst by future generations.
ROFLMAO.
Folks, we’re not there yet. Don’t count on cashing those universal basic income (UBI) checks that Sam wants us all to have once the machines do all of our thinking for us. ChatGPT is a truly stunning new technology, but it isn’t remotely intelligent or sentient.
Also, there is a real danger in the development of LLMs, although it's not the one that Lex and Sam would like us to believe.
My take?
The danger of LLMs isn’t that they may become sentient and decide that we humans of mere puppy intelligence are cute but irrelevant. The danger is that a small group of people will persuade us of this narrative, and we will accept it.
So what exactly is ChatGPT? It’s a massive statistical model of how actual people have used language over the years. It is fed billions of examples of historical texts, and it forms a predictive model of how to string words together to mimic these historical texts.
Is it super cool and revolutionary, and can it do some things much better than humans? Absolutely. Is it on the verge of surpassing human intelligence? Give me a break.
If you’re sweating a future with AGI running the show, you can take a deep breath and relax. GPT-4, the latest version of ChatGPT, is not AGI, and no one has any idea whatsoever what it will take for ChatGPT to achieve AGI. Heck, no one seems to have a definition of what AGI is other than, “We’ll know it when we see it.” How convenient.
That being said, we mere mortals do have some work to do. We have to up our game, because the best way to prove that ChatGPT is another great new tool is to use it!
I’ve started using it to help with my investment research. It’s an incredibly powerful tool, and it’s going to get more and more powerful every day. Moreover, it’s actually fun to use, and it helps me produce better results in less time.
The best way to interact with ChatGPT is to tell it who you want it to be and what you want it to do. Then carry on a conversation with it as if it were a first-year undergrad research assistant. For example, I prompted it as follows:
You are AdvisorGPT, and you're going to help me build my own personal securities portfolio that is well diversified (according to modern portfolio theory), while also prioritizing investments that reflect my personal values and interests. You are my research assistant. I look forward to working with you on this.
I’ve been carrying on a conversation for days with AdvisorGPT, and it keeps getting better and better. I’ve shared my investing thesis and even my current portfolio. I’ve told it that I like Ray Dalio’s holy grail portfolio, I’d like to invest in real things, and I’m concerned about inflation.
Over the course of the conversation, I’ve learned about new asset managers that ChatGPT suggested (Harding Loevner, Polen Capital, and Artisan Partners). It also identified a couple of potential investment ideas that could help get my existing portfolio closer to an inflation-proofed holy grail portfolio that’s invested in real things.
I even got it to outline the core elements of a holy grail portfolio – 15 to 20 good, uncorrelated return streams – and to give me ETFs for the asset classes along with allocation percentages for an “aggressive investor with a 10-year time horizon looking to invest for retirement.”
Ready for this?
1. US Equities (40%) – Large-Cap: SPY (20%), Mid-Cap: IJH (10%), Small-Cap: IWM (10%)
2. International Developed Market Equities (15%) – EFA
3. Emerging Market Equities (15%) – EEM
4. Sector-Specific Equities (15%) – Technology: XLK (4%), Healthcare: XLV (4%), Consumer Goods: XLY (3%); Utilities: XLU (2%), Industrials: XLI (2%)
5. Fixed Income (5%) – Government Bonds: GOVT (2%), Corporate Bonds: LQD (2%), Municipal Bonds: MUB (1%)
6. Real Estate Investment Trusts (5%) – VNQ
7. Commodities (5%) – Energy: DBE (2%), Precious Metals: GLTR (2%), Agriculture: DBA (1%)
8. Alternative Investments (3%) – Hedge Funds/Private Equity/Venture Capital: Not easily accessible via ETFs; consider mutual funds like the AQR Multi-Strategy Alternative Fund (ASAIX) or the Blackstone Alternative Multi-Strategy Fund (BXMIX)
9. Cryptocurrencies (1%) – GBTC or ETHE (note these are not ETFs but publicly traded trusts that hold cryptocurrencies)
10. Cash and Cash Equivalents (1%) – SHV
Umm … pretty cool, huh? Should you go invest in this portfolio without any further research? No way! But man, what a great leg up for getting started. And it pulled all this together in less than a minute, once I had provided all the information it required.
I highly encourage everyone to start learning about ChatGPT and other LLMs and start playing around with them. As with all great new technology, the best way to calm your fears is to start using it.
Things are still at a very early stage, and there are risks galore, especially big privacy risks. I would never share anything with ChatGPT, for example, that I wasn’t comfortable making public.
But whatever you do, don’t sit back on the couch and wait for your UBI check to arrive while Sam Altman ponders whether he’d like his final resting place to be Westminster Abbey or the Pantheon.
Get in the game.
Fight the noise,
Dr. Richard Smith
P.S. In my role with the Foundation for the Study of Cycles, I’ve been devoting a lot of my time lately to some exciting plans that I think will interest investors. Please sign up for the newsletter to find out more about what’s to come in 2023!