DEFINING HUMANITY
As machines increasingly assume human qualities (and if some humans enhance themselves to take on machinelike qualities), lines will become hazy. What is AI and what is human will change and, in some cases, merge. In coming to judge how we must keep pace with AI, humans will therefore need to assert more clearly what it is that distinguishes us from machines. How, then, will we compile and compress the entire range of human experience for easy comprehension by AI?
To preclude either our demotion or our replacement by machines, some will want to lay claim to difference via our proximity to divinity. Others will wish to ascertain more tactical conclusions: which kinds of decision-making can be delegated to machines and which cannot. We propose the articulation of an attribute, or set of attributes, that most of humanity could rally behind and orient around: one that will supply a floor underneath what is preferable but not a ceiling to what might be possible.
As a starting point, we would encourage a definition of dignity. Without a shared definition, we will be unable to reach agreement if and when AI is being used as a method or justification for the violation or erosion of dignity, and we would thus be hamstrung in our response. Without a definition of dignity, we would not know if and when AI, given enough faculties, could become a being of dignity, could stand fully in place of a human, or could be entirely unified with a human. An AI, even if sustainably proved to be not-human, might instead constitute a member of a separate, similarly dignified category that would nonetheless deserve its own, equal standard of treatment.
One conception of dignity, developed by the eighteenth-century philosopher Immanuel Kant, is centered on the inherent worth of the human subject as an autonomous actor, capable of moral reasoning, who must not be instrumentalized as a means to an end. Could AIs come to fulfill those requirements? A definition of dignity, we believe, would help humanity answer some of these questions and encourage inclusive coexistence with AI while avoiding reckless attempts at premature coevolution.
Both to retain an understanding of ourselves and to ensure that an appropriate conception of humanity is transmitted to machines as they learn, we humans will need to recommit ourselves to more than theoretical definitional work. The actuation of agency, curiosity, and freedom, renewing and exercising our inquisitiveness about other humans, about the natural world, about the universe, and about the possibility of the divine, will assist us in our ongoing participatory redefinition of the lines of humanity.
We will especially need to ensure that, beyond conventional ideas of worth like value and power, intrinsic human importance becomes one of the variables that defines machine decision-making. For instance, mathematical precision may not easily encompass the concept of mercy. Even to many humans, mercy is an inexplicable ideal, if not a miracle. For its part, and without considering the rules-based aspect, a mechanical intelligence might operate by valuing machine overperformance more highly than human performance. In such a situation, could the logic behind mercy, even if it cannot be formally taught, perhaps be absorbed? Again, dignity—the kernel from which mercy blooms—might serve here as part of the rules-based assumptions, or the iterative learning, of the machine.
Clear articulation of specific defining human attributes—particularly those that are, like dignity, widely integrated into both international political instruments and global faiths—could guide human efforts during periods of disorientation, including the choice between activity and passivity, the potential limits of self-evolution, and the precise transformation of AI in the direction of the human.
To illustrate the concept’s utility, take as a point of departure the following definition: Dignity is a quality that inheres in creatures who, born vulnerable and mortal and thus full of insecurity and fear, and despite their natural inclinations, can and do exercise their freedom not to follow their conception of evil but to choose their conception of good. In other words, those who can achieve dignity should, and those who do achieve it merit a special level of respect.
Undoubtedly, this definition is imperfect. It may leave out some living humans who are unable to make decisions—for instance, a person conscious but unresponsive—whom we nonetheless believe to be deserving of recognition as beings with dignity and thus entitled to respect. In this connection, perhaps the definition should be revised to indicate that dignity, once won, does not depart from us should we become unable to continue the actions that won it to begin with. A thousand such hypotheticals and emendations could be imagined.
Does this dictate our urging those who have chosen passivity in the face of powerful AI to exhibit agency and activity instead? If active, practical commitment is part of any moral ideal, then yes, the realization of dignity should be advanced. Under our definition, freedom is part of the human ideal, and thus we may expect—or even demand—that humans retain and exercise the power of conscious choice in the age of AI.
Under this definition, can AI itself possess dignity? Likely not—for AIs are not born, do not die, feel neither insecurity nor fear, and do not have natural inclinations or individuality such that conceptions of evil or good could be considered “theirs.” While AIs of the near future may appear otherwise, having personalities, expressing emotions, telling jokes, and recounting personal histories, under this framework they should be treated, philosophically, like literary characters. They may embody elements of humanity; they are not real in a moral sense.
Even the greatest literary character—Shakespeare’s Hamlet, for instance—is no more than a special combination of words, once written on a page and now reproduced many times over. “Hamlet” has no ability to feel a prickle in the eye, a turn of the stomach, a hot wave of frustration at thwarted expectation. “Hamlet” has no freedom to make a new choice. “Hamlet” is trapped inside his play. “Hamlet” is not a human, but a picture of a human. AI, made of strings of code and hunks of silicon, is much the same.
No doubt, some humans would therefore decry this definition of dignity as unhelpful, both philosophically and in substance. It might be criticized for being too low a common denominator—vague enough to appease all parties because of its excessive malleability—and concurrently for not capturing the idea that humans are worth preserving for their own sake and that, in some way, we are exceptional beyond our ability simply to survive. The philosopher Arthur Schopenhauer cursed dignity as “the shibboleth of all perplexed and empty-headed moralists.”14
But dignity, as we define it, usefully holds up our fragility and our potential for failure as well as our vitality, freedom, and ability to manifest our beliefs. It points at the good that we are capable of but have not achieved, and urgently, scoldingly, whispers: Go.
Still, admittedly, dignity alone cannot be enough. Other attributes should be specifically considered and perhaps added to the conception of humanity as it will be deployed in our coming partnership with AI. But our ability to define and sustain core elements of humanity as a baseline for AI’s comprehension of humanity at large is now a problem of existential significance, and the work to inculcate our definitions must be done now.
No definition will remain static; no doubt, as our own identities shift, we will need to evolve AI’s understandings in perpetuity. Meanwhile, others more capable than ourselves will continue to advance our collective thinking on the dynamic relationship between “us” and “them,” and their genius might yield a conception of humanity that more strongly (even if futilely) aims at ensuring our survival as a recognizable species. Even as they do, however, we should all strive for a definition, and a program, that goes further and elevates the human condition to new heights. For may not AI itself yet turn out to offer the strongest evidence of humanity’s ability to become an active participant in creation?
Comments
Post a Comment