AI has proved its power in many ways … but that doesn't mean we should give in blindly to all of the hype.
Back in 2016, the law firm of BakerHostetler became the first high-profile customer of a financial product sold by ROSS Intelligence. It was an AI, billed as “the world’s first artificially intelligent attorney,” whose job it was to help the firm with bankruptcies. Specifically, it did the job of a legal assistant, answering questions about financial laws by combing through all potential precedents in the blink of an eye and giving answers in language you didn’t need a computer-science degree to understand. Since then, machine learning has evolved to the point where even Hollywood screenwriters are worrying about being replaced by tireless electronic minds … and investors are getting financial advice on questions about when and how to buy and sell. “AI” has become a magic word, just like “dot com” was 20 years ago and “crypto” was five years ago. If you recall, both of those examples had highly publicized booms and equally newsworthy busts.
So how can we separate the hype from the value with AI?
AI In Brief
The premise behind artificial intelligence is that computers can be trained to understand questions and give answers in ordinary, human language but with all the speed, calculating power, and vast information storage of an internet-connected machine. Part of the trick has been creating a program that can mimic human activities (anything from turns of phrase to artistic styles) by analyzing very large libraries of data. That “data” can include, for instance, every movie, audition tape, and interview of a particular actor, or all of a particular writer’s published work.
Another way to describe the same thing is to look at the name of one of the most famous AI applications out now, the text-based ChatGPT. It chats like a human because it is a “Generative Pre-trained Transformer.” In other words, it Generates new words, sentences, and even whole articles because it’s been Pre-trained on millions or even billions of previously written texts, which it then Transforms by mixing, matching, and recombining elements. Everything is a jigsaw puzzle in this system, and the AI specializes in finding pieces that fit together, or at least seem to fit together.
On the other hand, people who work with AI talk about its capacity to “hallucinate” when the pieces are assembled in strange ways: making up names of reference books while writing papers, or putting grinning mouths in the place of eyes while generating portraits.
Evaluating AI
The hype swirling around AI — “It will disrupt every profession!” “It will end war!” “It will balance the world’s economy!” — is at best unhelpful and at worst misleading. But the technology is real, and is already making a difference. Some Wall Street trading desks, for instance, use AI for high-frequency trades, making investment decisions in milliseconds. AI is here to stay. That doesn't mean we should throw caution to the wind. When evaluating a stock or fund, don’t make a hasty choice based on it having the magic word “AI” in its name. Instead, think about what the technology actually does and what it’s likely to be doing in the near future.
Perhaps compare AI to the internet itself. In the 1990s, any company with the magic words “dot com” in its name enjoyed a fabulous amount of hype, it’s true — and veteran investors still talk ruefully about the dot-com boom, and the fallout after that bubble burst. The tech-heavy NASDAQ Composite Index soared from around 750 in 1990 up to over 5,000 in March 2000, then plummeted 78% by October 2002. Yet the words you’re reading now were researched and written on the internet, and some of the biggest shapers of our society are firms that solely exist on and for the internet. The “new economy” promised by early dot-com speculators is the reality we’re living in today. (And if you don’t believe that, just google it!)
When considering adding something AI-related to your portfolio, investigate what the technology is actually being used to do, and how that application is going to be used to actually increase revenues over costs. Do your best to look for those costs in unexpected places.
For instance, it might seem like an AI, as a kind of computer program, only needs a sufficiently advanced computer and a staff of coders to keep it running. But computers need electricity, and when they’re working hard, they tend to get hot. A recent study at University of Massachusetts Amherst found that training a single generative AI model can consume more than 75,000 gallons of water. That’s as much water as an average person drinks over the course of 27 years, just used to keep the machine cool while it’s absorbing all its data.
Then, once the model is plugged in and running, there are more factors to consider. The technology has shown a vast capability to mutate, teaching itself new tricks and offering new potential at every turn. Is the firm or market sector you’re evaluating flexible enough to take advantage of new directions, but stable enough not to lose sight of their original mission? The models also show a capacity to ignore the usual boundaries around protected sets of data. That is, AI is smart, but not smart enough to understand concepts like “copyright” or “privacy.” For the firm or market sector under consideration, is there enough human oversight to avoid legal and ethical problems that arise? Think about the recent Hollywood writers’ strike for just one example of an unexpected revenue stream (replacing the pros who rewrite screenplays to order) and an equally unexpected quandary (an industry segment up in arms, shutting down TV and movie production for months).
As always, it’s important to keep a balanced, informed view of potential risks along with the promise of new developments.
If you have more detailed questions about the ins and outs of investing for your goals, ask one of our financial advisors. They’re more than willing to provide answers … from a live person.
Sources:
* https://www.coindesk.com/consensus-magazine/2023/06/21/crypto-and-ai-save-us-from-the-hype
* AI bankruptcy attorney: https://www.law.com/almID/1202757054564/
* https://www.investopedia.com/financial-advisor/how-ai-shaping-advisory-landscape/
* What GPT stands for: https://www.advisorengine.com/action-magazine/articles/artificial-intelligence-is-a-game-changer-for-financial-advisors
From Investopedia on dot-com bubble:
* U.S. Government Printing Office. "Financial Crisis Inquiry Commission Report," Page 59.
* U.S. Securities and Exchange Commission. "CFA Institute: Re: Study Required by Section 989G(b) of the Dodd-Frank Act Regarding Compliance with Section 404(b) of the Sarbanes-Oxley Act (File No. S7-29-10)," Page 2.
* Environmental costs of AI: https://www.schroders.com/en-us/us/individual/insights/ai-revolution-what-s-the-environmental-impact-/
* Forbes on hidden costs: https://www.forbes.com/sites/forbestechcouncil/2023/08/31/the-hidden-costs-of-implementing-ai-in-enterprise/?sh=2dc088ab4d5c