Today’s hot topic is artificial intelligence and what it can do to shape and change the way we do business. Its ability to analyze vast amounts of data and make predictions based on that analysis has led many to believe that it can help us formulate the perfect business strategy. But can AI-based agents become your new chief strategy officer? To answer this question let’s consider what we can and cannot do with AI when it comes to strategy formulation.
AI-powered tools (such as ChatGPT, to stick with the most famous) are, strictly speaking, predictive language models. Sifting through terabytes of data, they can provide plausible responses to specific questions, simply based on statistical and probabilistic techniques to determine the most likely sequence of words occurring in a sentence used to answer a specific prompt or query. No matter how good those answers seem to be, we should always remember that they come from the most statistically frequent sequence of words associated with one specific question. In other words, tools like ChatGPT generate language, not original thinking.
The second thing we should always remember is that ChatGPT uses as sources what is on the web. And while there are a lot of good things on the web, it is also full of dubious sources. Unfortunately, ChatGPT takes the information in a scientific paper published in a reputable peer-reviewed journal and puts it on the same level as information from a less-than-reputable blog or website. So, as computationally sophisticated as ChatGPT is, it is only as good as the sources it uses.
The third thing that should give us pause when thinking about delegating strategy development to AI is the fact that with strategy we rarely deal with clear-cut, structured problems. Formulating a strategy requires instead dealing with issues that are broad in nature, ill-defined, and that can possibly be addressed in many different ways. In fact, it is often said that with strategy, being able to ask the right question is more than half the battle. The perfect answer to the “wrong” question could translate into a massive waste of resources due to misplaced focus; this would not help much in devising a successful strategy. So, while ChatGPT answers prompts, it is not able to come up with the right questions.
For all the amazing features this technology has, it lacks the most fundamental one for the successful formulation of a business strategy: Informed Judgement. Judgement is defined by the Cambridge dictionary as “the ability to make good decisions or to form sensible opinions about something,” and informed as “having knowledge or information about something.” This is why even this technology, as advanced as it might be, will not substitute the human ability to think critically, to appraise and evaluate sources, to assign priorities and to decide by blending idiosyncratic experiences, objective data, and contextual knowledge and information. That remains entirely a human prerogative.
So what can be good about AI when it comes to strategy?
First of all, by being able to process large amounts of data in a short amount of time, AI allows you to identify patterns that might not be immediately visible or obvious from less structured or systematic observation. In some industries such as finance, healthcare or retail, data processing and forecasting are very important to make informed decisions, particularly to spot emerging trends or recent changes that are just starting to appear.
Second, by not filtering information or sources, AI could help us search with limited effort areas that we would normally ignore, given time and resource constraints. In fact, since searches are costly, often when seeking new solutions we tend to be conservative in our approach; we only look for solutions to specific business problems in a radius that is quite limited. We’re a little bit like the proverbial drunk who has lost his keys at night and only looks for them under the lamppost since it’s too dark to see anything anywhere else. AI can provide an efficient scan of more distant, less obvious options.
Finally, we could use AI’s computational capabilities to run different scenarios in response to the strategy we are considering. Traditionally in business strategy, we explore the implications of intended moves or decisions against the backdrop of different scenarios, but we do this mostly as thought experiments, which means we tend not to be very analytical about all the possible implications. AI can easily project different scenarios based on the parameters we select, and although they might not be 100% accurate or precise, these scenarios could help us identify blind spots in our strategy that normally we would miss.
It should be very clear from the previous discussion that the best use of artificial intelligence is as a complement to and not as a substitute for human intelligence and ability. As with all technological revolutions, the differences in the future will not be between humans and AI, but between humans who know how to use AI and humans who don’t. In that sense it is important to learn the potential of this tool, what it’s good for and what it’s not, and most of all to use it to enhance, not supplant, what humans can do.
To close, one of the main tenets of business strategy is that a good strategy should help you “beat the market.” Beating the market means doing better than the average player in your market. If you are simply average, it is not a matter of if you will be selected out of that market, but how long it will take. Survival requires you to be better than average.
Currently, given the way it works, given the sources it uses, given the statistical / probabilistic nature of the algorithm, AI tools such as ChatGPT, at best, will put you at about average in a hypothetical distribution of success in a given market. Human judgement instead, remains the most powerful tool to help you get ahead of competition.