Are Your Money Managers Chatting With Bots?

Hype and Reality—I assemble, therefore, am I?

The last few weeks of ChatGPT/Bing vs Bard chatter online and in the press has elicited an outpouring of commentary on AI.

Everything from “Sydney” the would-be home-breaker, to evil alter egos, to espionage on the AI’s creators has graced the media.

Beyond the hype and disparaging criticism where do investors look to find where the real opportunity in AI lies in helping manage any of their investment portfolio? If AI is harbored by traditional tech, is there really much further opportunity to be extracted at the investment portfolio level?

The sudden overt sparing between Microsoft and Alphabet in recent weeks over who is leading on AI may be interesting fodder for the media but what realistically will the impact be on a portfolio, or for portfolio management? Many investors hold stock in both Alphabet and Microsoft, often somewhat proportionate to their relative market capitalization; $1.17 Tn (GOOG) and $1.85 Tn (MSFT). Therefore, so what If ChatGPT ultimately gives Microsoft the jump on Google’s Bard?

If someone has a moment of idleness, he or she could ask the all-new Bing to find some human-like answers to all of the above.

Part of the issue in finding clarity is confusion over the idea of intelligence. For much of the last 150 years an educational mindset has developed, and settled upon, a meritorious process revolving around grading and examinations which, at core, rewarded memory skills to the detriment of understanding. Hence intelligence gradually became synonymous with knowledge which essentially came down to having good recall ability.

AI in its ChatGPT form is superhuman memory that some, in a confused sense, term as intelligence. However, this confusion is human in origin and human in interpretation. The all-new Bing’s current catch line is “Ask real questions. Get complete answers. Chat create.” So, let’s break down both the semantics and the realities:

Ask real questions: This exemplifies the enormous advance in natural language processing (NLP) with the success of converting human language into machine readable form. Note this does not mean that the machine understands anything but rather that it can ascribe it to the correct analytic bin.

Get complete answers: This further reflects the advancement of NLP and references a more complex stage than the asking of questions. It is the ability to analyze and process very rapidly volumes of human information through the correct analytical prism, to which the initial human question was correctly assigned. Next, elements from different sources are assembled and stitched into a form of language output that is human-like in structure and nuance.

Do not be fooled by the flood of disparaging commentaries that hinge on catching the AI out, by pointing out its vocabulary or point of reference mistakes and errors. These commentaries mostly fail to acknowledge the enormous technical achievement that it is to have succeeded in bringing NLP to the level that it has reached. Furthermore, every such error is learned from, which explains why in some cases, subsequent attempts by some to repeat the catching out do not work.

Chat create: Well, no actually. Think of it like an advance on a basic digital calculator. Ask real questions “2×2=?”, get complete answers “4”. It is unlikely that one would describe that kind of interaction as having a chat with your calculator. Similarly, “Chat” with the AI “Chatbot” is not chat, but it certainly is input and output.

As far as creation goes, there is the creation of a larger database of human input — machine output verifications for Microsoft, which will allow the machine output to improve over time. Perhaps also there is the creation of the illusion that one is having a chat with an intelligent machine and if one goes full emersion in delusion, that it is a chat with a machine that is a potentially sentient being.

As an investing tool, is ‘Chat-type’ AI that’s ready for prime time?

Wall Street appears to already have turned up its nose to the “potential” of chatbot analysis, as reported just days ago in a Bloomberg article.ⁱ It is easy to understand why. Fears of inaccuracy or the inability to replicate outputs in a consistent manner make for a regulatory and compliance horror-show. From a Wall Street compliance perspective FOMO (Fear of missing out) can quickly and overwhelmingly be supplanted by FOBB (Fear of being burned).

The article noted that “traders and other professionals that are dabbling in the technology are quickly finding that, while it could make some of their most mundane tasks faster, the process is hardly seamless.” Bloomberg went on to quote Oded Netzer, a professor at Columbia Business School who researches data and technology, who said “It may save time, but we don’t know if it’s true, which is the biggest downside of the tool.” As the article went on to share, ChatGPT struggles with math.

A comment in the article by analyst Larry Tabb at Bloomberg Intelligence got to the nub of the issue: “When the SEC knocks on your door and asks why did you execute that transaction, you have to have a better answer than, ‘Well the machine told me to,’” he said. “You can gather insights and analysis from AI and then program your computers, but black-box trading models are generally not sanctioned on Wall Street because you don’t know why the trading decision was made.”

If not the flashy ‘Chat’ AI for investment management purposes, then what?

Money management is deeply nuanced, and error is profoundly consequential. This is a point which we at Plotinus have stressed time and again, as anyone familiar with reading our commentaries on the use of AI in investing will know. We believe that proper usage of AI in investment decisions means that the humans behind the technology must be doing the thinking and deeply understand both nuance and error.

For investors looking for the real advantage with AI for incorporating it into running an investment portfolio, in our opinion it lies in a less flashy, less hyped field: AI-based trade decision-making. There is nothing certain about the future, so AI trade decision-making is dealing with an unknown. This could not be more different from the excellent attempts that the AI chatbots represent, in their pursuit to locate relevant, prior known information and assemble it a form that a human can comprehend. Yet it is in this field of investment trade decision-making that the benefits of the super/non-human analytic capabilities of AI are of the most potential use for investors. Deploying AI as tool can help attempt to answer unknown questions, by assessing market directionality and the like. When this is done in the context of a human understanding—translated though a clear investment methodology and process—it becomes a very powerful investment application.

We at Plotinus are real humans and we are more than happy to “chat” personally with investors interested learning more about what AI-based trade decision-making can bring to create a better overall investment approach and the potential to generate enhanced investment returns. ■

ⁱ Doherty,K. (2023, February 24). Wall Street’s ChatGPT Nightmare Is Over Before It Starts as Banks Crack Down. Bloomberg.