The Coming of the Age of AI
As we prepare to enter the third decade of the 21st century, we’re entering into a brave new world flush with unparalleled amounts of and access to data. I see many striking implications coming from this. On the less severe side, companies have been able to determine that a teenager was pregnant before her father could by mining shopping data. Darker applications include the use of data by companies like Palantir to profile individuals based on their predicted probability to commit crimes (just as in the movie Minority Report), or by governments such as China to enforce a social credit system that extrapolates from one’s daily actions to evaluate and predict how “good” of a citizen one is.
Indeed, these applications can have benefits as well as disadvantages. For example, if data indicates that one has a high probability of committing a crime in the near future or that one is a bad citizen (by some government-determined metric), then institutions may proceed to act on the suggestions of the data by limiting the freedoms of the individual in question. This may lead to a safer, more productive society — yet, at the same time, this approach can lead to a slippery slope where data is used as a justification for the curtailing of individual liberties.
This is the world we’re entering. This is our future. Tomorrow’s generations, from the Millennials onward, will live in this world. It can be dystopian — the stuff of science fiction stories — and the fear of a data-driven future has only been stoked by the channels of modern mass media.
But a data-driven society can also be seen as the pinnacle of a search for knowledge and truth that has lasted from the very beginnings of human society. Much of human history is the story of people trying to create ways to reduce uncertainty and make better decisions — in essence, to find truth. The pursuit of knowledge has benefited our human race greatly — no longer do we sit by cluelessly as a Black Plague kills a third of a continent as our medieval ancestors did, and no longer do we perish as a result of eating the wrong berry as our hunter-gatherer forebears did.
Yet, we still have much room for improvement. Decision making is still imperfect. But it doesn’t have to be. In theory, in every set of decisions exists an optimal choice. With big data, it isn’t so farfetched to imagine a world where the optimal decision can always be determined. For the first time, with our new access to hordes of data, the ultimate truth can be pinpointed more precisely than ever. Truth can prevail and serve as the basis for all decision making, and we’ll finally be able to make perfect decisions.
With the advent of big data and machine learning, the probability that an outcome will happen, whether that be if a woman becomes pregnant or if a person will commit a crime, can feasibly be predicted with high accuracy in the near future. And once we reach the end of that search for truth, then we’ll be the most informed people to ever live. In every case, an optimal decision can be found — there will be no more uncertainty about whether or not a choice is the right one. Data can determine for certain what’s best.
But if we envision a world where we can know for certain what’s optimal, then, in such a world, our ability to make suboptimal decisions will be taken away. In other words, free will and autonomy will be taken away as we all follow the predetermined optimal decisions that have been laid out for us by the data.
But if free will is simply a means to achieve the end of good outcomes for ourselves, then is there a need for the means of free will once we find foolproof and accurate ways to achieve the end of good outcomes? Certainty relieves us of our free will, which, by its nature, relies on chance to present us with the illusion of choice. When we’re uncertain of the outcome of our will, the pressure overwhelms the illusion.
And without uncertainty and free will, adversity is impossible. If every decision can be made optimally without any mistakes, then it will be highly unlikely that adverse situations can arise. Data-driven optimality could completely relieve humans of stress, inconvenience, and missteps, replacing them instead with luxury and comfort.
Is there value in adversity? Does it enhance life in some way, perhaps by enriching experience or contributing to individual growth? Is it purely a negative aspect of life that we’d be better without? And if the answer is the former, then how would individuals in a society that lacks exposure to adversity behave? If there’s value in adversity, then is there also value in making suboptimal decisions?
These questions are important to directionally guide the application of big data technologies. It determines what these applications should optimize for. If the goal is perfect information, ultimate efficiency, and optimality, then we may just as well end up in a world of luxury and comfort, and lacking in adversity. We could create a hedonistic utopia where pleasure is maximized and pain is minimized. In many ways, one can see the history of mankind up to today as a pursuit of that very goal — maximization of pleasure and minimization of pain. Would such a thing be desirable? At first glance, it seems like the answer is yes. More of a good thing (pleasure) and less of a bad thing (pain) must obviously be good on net. But the question is more complicated than that, and this work seeks to develop an adequately nuanced answer.
Some other questions to be explored are as follows.
- What exactly is the nature of free will?
- What’s the use of free will?
- What’s the value in suboptimality and adversity, if there is any?
- What does it mean to be human?
- What should be the goal of society?
These questions have driven me to create this work. To better answer these questions, I’ve drawn upon my background in statistical science and spoken to many experts in the Artificial Intelligence and Machine Learning fields.
To explore them, I’ve created a world with a data-driven machine that can optimally decide the best decision for any and every fork in the road. One would never make a wrong move ever again; one would only make the moves that the machine tells them to make. What would it be like if such a machine existed?
n my view, free will is the child of uncertainty, and a world without uncertainty is a world without free will. But is that so bad? For a world without uncertainty also means a world without suboptimality.
This book is for those of us who are interested in the philosophical implications of data, who enjoy ruminating on the nature of free will, and who are on the precipice of this data-driven utopia/dystopia — all of us.