The Dog Days of Summer and AI

​I usually speak about the summer doldrums this time of year, but I’m starting to like the dog days of summer better. As always, our phone stops ringing other than the four or five daily robo-calls we get alarming us about our Google listing. Total market trading activity slows down, and people either go on vacation, or find it hard to get things done while thinking about going on vacation. A recent article from CNBC talked about the 4th of July week being the most traveled week of the year, but also the least productive in the workplace. A lot of the younger generations take “quiet vacations” that week, which I think means they take vacation without asking for paid time off.

​After looking into the history of the dog days of summer phrase, I find it a much more fitting description of this time of year. I made it easy on myself and used Wikipedia, and learned that the phrase goes back to ancient Greek times. It describes the time of year when Sirius, the Dog Star, undergoes its heliacal rising in the Norther Hemisphere. That just means the time when the star first becomes noticeable in the dawn sky, shortly before sunrise. The Greeks, along with several other cultures, noticed this time as being wrought with excessive heat, drought, thunderstorms, lethargy, fever, mad dogs, and bad luck. I believe many of us are able to tick a few of those off on our fingers as something we’ve been experiencing lately.
 
Regardless, the summer months offer opportunities for vacations, a good amount of time to think, and for those who are extremely motivated, the chance to get a little extra done at home or in the shop. Craig recently went on a fishing trip to Canada with his son and grandsons, which had been delayed from Covid times. One of the benefits of the delay, however, was an additional grandson to vacation with! His trip was a little wiser than many of our summer vacation targets are… He went north, into an area with a much nicer temperature. Kendall recently went on a roughly 2,000-mile jaunt on his motorcycle, ending up as far north as St. Johnsbury, Vermont. Still, much of his riding was amidst brutal heat. As for me, Robyn, Carter, and I will be going to Myrtle Beach in a few weeks, to punish ourselves under the sun. Despite the heat, I am looking forward to it.
 
While I do spend my fair share of time daydreaming about vacationing, I also try to fit in some thinking as well. What has been pinging my radar recently has been all things artificial intelligence and technology. As I’ve confessed many times now, I have always been a computer nerd, starting at a very young age, as well as a fan of science fiction. So, all the hubbub and change related to artificial intelligence fascinates me.
 
I process a lot of what I read about AI into how it will affect investing, though I also enjoy reading and thinking about how AI does and will affect our everyday lives. I am a big fan of Bojangles’ use of AI to take my order at the drive-thru. We both understand one another, the AI is a stickler for details (no mayo!), and the interaction advances at a favorable and efficient pace, although occasionally the AI incorrectly says they are out of honey-mustard. The drive-thru attendant offers a correction at the window.
 
For those of you willing to go down a certain rabbit hole, look up the stir that romantic relationships with AI have caused in the world. This isn’t supposed to be off color - I just find it shocking in how whether it’s an AI drive-thru, telephone customer service rep, or actual boyfriend or girlfriend, many people find it easier to relate to an artificial intelligence than a real person.
 
Artificial Intelligence has created lots of ripples in the investing industry. Obviously, all things AI related have been targets for investment by everyone from individuals to hedge funds, with NVIDIA being one of the most prominent targets lately. I am a little concerned that, like with the dot-com bubble, people might be convinced to invest their savings in a no-name startup tech company whose only business plan is “a goal to be the next biggest AI firm.” More worrisome though, is the “revolutionary, never-before-seen, AI-enhanced, guaranteed to outperform investment models” that will be sold to people.
 
Computer driven investment models have actually been around since the 1970s. Dad himself wrote many times about the “Quants,” or investment managers who invest their clients’ savings using quantitative computer algorithms, and the risks and failures they’ve created in the past (Long-Term Capital Management). The technology that drives these algorithms has grown more complex and robust, though they can offer certain benefits of analysis to managers and retail investors.
 
However, despite the prolific advancement of our technology, especially in computer data storage, processing, and analysis, the fields of economics and finance remain better defined as social sciences. They offer us insights into the interactions (working, spending, saving, buying, holding, selling, giving, and taking) of economies and capital markets, but those insights are solely dependent upon an understanding of the psychology and behavior of their participants… people. The laboratory of economics and finance is overflowing with emotion and illogical human error.
 
Computer only investment programs that remove human oversight and control have never been able to make perfect investments, much less protect investors from the mass hysteria people are subject to. Also, common sense dictates that if someone were to create a truly successful program, they would have zero incentive to “sell” it to others. This would take away their advantage, as once it was incorporated by enough people, it would limit the profits available to zero, resulting in no outperformance for anyone.
 
Even the best Large Language Model Artificial Intelligences, such as OpenAI’s ChatGPT, ChatGPT GPT-4, and Bard, currently offer poor practical financial advice. A study described in the Journal of Financial Planning gives us a great example. In “LLMs Can’t Be Trusted for Financial Advice,” by Gary Smith, 15 separate prompted financial-decision questions were asked of each of those top LLM AIs. The questions involved things like which loan on a car or a home would be better, what the value of a life-insurance policy was, or when someone should begin social security, but with specifics as to amounts, ages, risk tolerances, etc.
 
Each AI gave answers to the questions that were “authoritative” and “grammatically correct.” They cited known rules, best practices, and calculation methods. The answers looked good and were confident. The only problem was that not a single AI gave a correct answer to any of the questions presented to them. Many of the errors were calculation errors, which should be the least likely area for mistakes to be made, though many of the mistakes were also practical ones that a human would likely easily spot using common sense.
 
It’s good to be open to new things, but it makes sense to be leery as well. We experience life through a human perspective. Many of the things we value most, especially our safety and health, can only be respected through that human experience. As the growing evolution of our technology creates widening space between people, anchor yourself in that human experience. All of us share an understanding of value through a human lens, and those spaces created by our technology are never greater than we, as people, can reach across.
 
Enjoy the Summer, stay out of the heat, and call us or visit when you can.
                                   
Justin Anderson