The earliest forms of AI were invented by Soviet Systems Theorists and Cyberneticists in order to address the problem of modeling complex chaotic systems. This was a vital missing piece at the core of Soviet Communism; the ideology was built around central planning of things like ecosystems and economies, and these forms of modeling did not work. This is why AI was originally invented, in order to create a system that could learn how to model complex chaotic systems and provide insight and answers where past attempts at similar analyses by humans had failed.
Imagine for example a complex social interaction between tens of thousands of people who share little else in common besides being members of a particular facebook group. This group is dedicated to criticism of a specific social problem. In this case, money in society and its constellation of perverse incentives and negative externalities. If tens of thousands of people agree that money in society is a problem, that does not necessarily mean that any two of those people agree on the solution.
The task of AI becomes the task of defining what such a group could be. It’s not just a question of modeling the group and its myriad opinions and perspectives but finding a way to push all of that chaos towards some convergence that becomes capable of arguing for a solution to the problem the group exists to discuss and critique. In many ways, this has become a central issue of the modern era; everyone has opinions about the problems in society, and there are as many opinions as there are people, with very little overlap between arguments and conclusions derived from myriad disparate sets of experience.
The temptation, and the precedent throughout history, has been to elevate one perspective or one figurehead to represent one set of experiences and arguments and conclusions and privilege those above all others, forcing everyone to adopt policies and practices that represent this limited perspective as a universal solution which in reality fits only a limited subset of experiences.
A naïve perspective would argue that the challenge of AI is to create a system that can take in all of these disparate perspectives and find a way to meaningfully synthesize them into something that can provide direction and insight where past attempts at such analysis by humans have failed. But this is just more of the same. A true synthesis of all these perspectives is not possible, and would likely be just as limited in its ability to produce direction and insight as any single perspective. The challenge of AI is not to create some grand synthesis of all human perspectives, but rather to find a way to create models that can learn from history and from the myriad perspectives and experiences of people in the present in order to provide direction and insight where past attempts at such analysis by humans have failed.
The task of the moment is to create a society which allows for the expression of a diversity of perspectives while also creating the conditions for people to learn from and build on the perspectives of others. Fundamentally, scarcity is the enemy, and scarcity in the modern world is a policy choice; there is already plenty of food and housing for everyone, but it’s being deliberately withheld from those who need it in order to force them to comply with some limited, external perspective on addressing their problems. A more just society would allow a diverse range of perspectives, experiences, conclusions, and arguments to coexist with a basic guarantee of survival for everyone who intends to coexist. In such a society, the only people who would not be welcome are those disagree with the basic premise that everyone should be able to exist and subsist and to pursue and contribute to their own liberation and that of their community.