I start this piece with some disclaimers. Firstly, this piece was not written by prompting artificial intelligence (AI) and the ideas contained therein therefore reflects my natural intelligence (or lack thereof, depending on your point of view). Secondly, I am not about to lead up to a conclusion that, although AI may replace certain jobs, many new jobs will be created, even some which we have not yet thought of. Finally, I will not be sharing thoughts on how an investment portfolio might be constructed to benefit from increasing adoption of AI. Each of us has already read many newspaper articles, opinion pieces and research views where the big reveal basically reflects one or more of the above. It has been done. The novelty factor has worn off.
Instead, I plan to share some of my recent musings on what AI represents in the bigger context of humans seeking to make better decisions. Granted, these might be less original than I would like to think, particularly as I am far from an AI expert, but are hopefully more interesting than the current proliferation of consensus opinions (frequently written by AI). Coincidentally (or maybe not), two important points are highlighted in the preceding sentence:
The quest for more information. For millennia, access to more information correlated strongly with making better decisions. Information was scarce and expensive. Putting in the effort to acquire just a little more knowledge was enough to secure an advantage over your competitors. Even today, this belief continues to reflect in a relatively universal feeling of admiration that we experience when entering a large library or an exceptionally well-stocked bookstore. Another manifestation of the same phenomenon is the innate respect that society affords to highly qualified individuals. Although we are aware that knowledge ages faster than ever before, we continue to place value on the time and effort invested to obtain the larger volume of knowledge associated with an advanced degree.
However, with the advent of the internet and the information age, information has become cheap and plentiful. In addition, the internet is democratic and does not discriminate between low quality and high quality information. As a result, we are overwhelmed by mountains of facts, ideas, opinions and trends. Nevertheless, the conditioning of millennia leads us to believe that we will make better decisions by consuming more and more information. In the information age, this belief often causes harm. Research shows that when we are faced with a larger number of options, we make worse decisions, mainly because we are more aware of the potential for regret. In this context, the rise of AI (as a tool to sift large amounts of information rapidly) reflects our attempt to deal with a world where the correlation between more information and better information has weakened.
Proliferation of consensus opinions. The AI that is currently in use uses statistical probabilities to sort existing information. For example, while ChatGPT has become part of the public consciousness because of its eerily human-like responses to questions, experts are at pains to point out that the technology relies on probabilities to select the words to construct an answer. In other words, given a set of words, it selects the next word based on what is statistically most likely to follow. This is similar to suggestions that we see when typing an email or a message on WhatsApp, just far more sophisticated. The ability of AI to assess statistical probabilities is already being used to concentrate human effort on the most promising possibilities. For example, medical scientists have used AI to narrow down the number of potential compounds when developing new treatments.
However, statistical probabilities have their limitations. For example, even an avenue of investigation with a very low probability of success could still be spectacularly successful. Many inventions that are in wide use today were discovered while pursuing other, sometimes unrelated, goals. Stainless steel, post-it notes and penicillin were all accidental discoveries, sometimes not solving the problem that was being investigated. In an investment context, those that chose to rely heavily on AI in investments, have thus far paid the price when events with low probabilities (but high impact) materialised. This is partly because statistical probability is heavily reliant on existing information (i.e. what we already know and what has happened in the past). Moreover, a plausible argument can be made that, as AI increasingly creates new information that is made publicly available, AI will increasingly rely on the information previously created by AI. This could lead to a situation where consensus opinions are continually reinforced within a narrowing discourse. If this should come to pass, dissenting opinions and truly new patterns of thought will become extremely rare and therefore extremely valuable.
In combination, the above suggests some conclusions. Humans are still trying to deal with the transition from a world of scarce information to a world of abundant information. Decision-making habits and beliefs developed over millennia are not as effective as they used to be and we have to determine how many (if any) of these are still useful. Developing ever more sophisticated AI is our attempt to deal with a world of plentiful information. Nevertheless, in many ways, AI still reflects a belief that processing more information more rapidly inevitably leads to better decisions. Slowly, we are beginning to realise the importance of better information, rather than more information. However, current AI distinguishes the quality of information based on statistical probability, which brings its own weaknesses to the sifting of information. One potential outcome of statistical probability is strengthening consensus ideas such as, ahem, the idea that, although AI may replace certain jobs, many new jobs will be created.
Applying the thoughts to our daily lives implies that we need to be conscious of our information addiction and how this could affect our decision-making and overall well-being. While AI is a useful tool to process large volumes of information, we might do well to remember that more information does not necessarily mean we are accessing the best information. In fact, the best information for a decision can be simple and superficially unimpressive. For example, simple investment models that accurately reflect business drivers can be worth more than complex models seeking accurate valuation. A quick response to a signal of significant weakness is often worth more than an accurate assessment of what the size of the weakness is likely to be. In other words, it is better to realise and act on simple information that a share price is likely to drop by around 50% than to spend time accurately modelling that the drop will be 48.562% before making a decision. In a different context, we might do well to remind ourselves that greater choice generates stress, as the chances of making the wrong decision appear greater. Practically, it means that we can increase our levels of peace and contentment by eschewing online shopping in favour of visiting our local mall and selecting from what is available. Or perhaps we should worry less about whether “warm sand” or “mushroom cream” is the right colour paint to select and admit that both are basically “beige”. Furthermore, it seems worthwhile to sometimes take a risk by exploring a less-travelled path. As we are increasingly guided towards consensus opinions and statistical probabilities, those of us who are willing to potentially waste our time on the unlikely face far less competition and increased rewards if we happen to make a worthwhile discovery. Such a decision requires substantial fortitude, given the low probability of success, but creates the best opportunity for natural intelligence to shine.
In summary, AI has existed for some time, has impacted the way we live and will continue to do so in the future. However, to get the most of the new AI iterations, we should remain aware of blind spots both in the technology and within ourselves.
[Another disclaimer: despite the reasons for doing so, I do not think I have evolved sufficiently to give up online shopping yet…]