Artificial Intelligence (AI) is accelerating everything that is process-driven and governed by clear rules. Due to its heavy processing power capabilities, AI can write code faster than anyone, calculate or solve a problem almost instantly, or suggest a diagnosis on a medical condition as long as the rules are clearly defined (mathematics, programming languages, etc.) and the “chain of causation” is clear and can be traced by the model.
Understanding the Chain of Causation
The causal chain, or chain of causation, is the relationship between a Problem, its Cause, and its Effect, or consequences.
It usually will look like this:
There is a Problem: “I need to pull up a chart of IBM stock prices for a year, but I don’t know how to code.”
The cause: a lack of Python coding skills.
The consequence: no outcome or result.
Because the cause (in this case: the lack of Python coding skills) is very clear and Python has a perfectly explainable codebase and set of instructions that leave no “interpretation” or wiggle room, the AI bot can easily generate the correct code within seconds.
The same logic applies to medicine. An AI model can diagnose a fever in someone with a “blocked nose, temperature, headache, and fatigue.” Again, the causation is clear, but there is more room for interpretation or variation, as another underlying condition in the patient could explain all of these symptoms better. This is where AI begins to encounter its limitations as “human” traits and behaviors come into play.
The Model of Language and Probability
Summarizing a text seems like a clear and easy task for AI models. Or is it?
AI does not understand language the way humans do. Instead, it generated words based on statistical probabilities. This works when there is a high probability that a certain word sequence makes sense, but it’s not when it drops to an abnormally low level.
To the human eye, it looks like intelligence, but in reality, it’s mathematics. The rules for solving the Problem-Cause-Effect chain are simple at first, as it is based on the English language and a plethora of sources on the web. However, the “sources” used are not always reliable and can be riddled with errors, biases, and misinformation. The English language also has subtleties, with words that carry emotional or behavioral meanings that are not recognized by the model.
So, while AI can efficiently summarize factual information, like news and events, it often struggles when tone, sentiment, or human intent are involved.
AI in Financial Decisions
Financial decision-making may seem like the perfect use for AI. You might think you can feed all the financial and economic data on companies into the model and simply ask it to “pick the best stocks” for you. However, it’s not that simple.
The rules of causation in finance are not always clear. Is a low P/E ratio better or worse? Is a highly leveraged company entering a new market better than a debt-free one in an established space? These types of questions don’t just involve numbers, but also judgment and strategy.
Chances are, an AI model that was out two decades ago would have likely missed Amazon’sstock in 1999, or even in 2005, when it was a leveraged, unprofitable business that turned into something big. Human investors who understood Jeff Bezos’ strategy, market disruption, and consumer behavior saw an opportunity that AI could not provide due to unclear rules of causation.
How Is AI Revolutionizing the Stock Picking Game?
AI is transforming the stock-picking process, but only under the right conditions, like data quality. With a clean, reliable source of data that minimizes the risk of error, AI can summarize news faster than anyone, process SEC filings, analyze conference call transcripts, and review other official documents almost instantly.
It can then group the results by themes or issues and present insights far faster than a team of analysts can.
AI can also sift through a vast database of financial numbers and ratios to identify outliers, abnormal levels, or unusual patterns, and surface a list of stocks that might be worth further reviewing.
And that’s where its role ends.
The rest, human biases, emotions, intuition, and crowd behaviors cannot be reduced to an equation.
That’s why any app claiming that “AI selects stocks for you” or “AI does it, no need to think” should be treated with skepticism. AI helps tremendously with time-consuming number crunching and text reading, but only if the data pool has been properly vetted, legally purchased or licensed, and ring-fenced to avoid errors and confusion. The rest should be left to humans.
Overall, AI is transforming financial decision-making, especially in areas such as stock picking, by making the process faster, more efficient, and more accurate. But it’s definitely not smarter, as Human Intelligence (HI) is required for that.
