Persons without Qualities: Algorithms, AI, and the Reshaping of Ourselves

Abstract

Persons without Qualities:Algorithms, AI, and the Reshaping of Ourselves Joseph E. Davis (bio) and Paul Scherz (bio) an elegant couple walking down a busy thoroughfare comes across the scene of a traffic accident. A man, apparently dead, has been struck by a lorry. Seeing him through the crowd lying motionless on the ground, the woman experiences an "irresolute, paralyzing sensation." Her companion observes that such heavy trucks have too long a braking distance. Oddly, this explanation is comforting: "Somehow the lady felt relieved at hearing this … [S]he did not know what a braking-distance was, nor had she any wish to know; it was sufficient for her that by this means the horrible happening could be fitted into some kind of pattern, so becoming a technical problem that no longer directly concerned her." An ambulance arrives with a satisfying promptness, and men in uniform quickly whisk the casualty away. The gentleman continues his reassurance, quoting American road accident statistics. The lady feels better, yet she can't quite shake the "unjustified feeling that she had experienced something exceptional" (Musil 1953, 5–6). In this vignette, from the first volume of Robert Musil's masterwork, The Man without Qualities, we glimpse the strange power that we explore in this special issue on algorithms, artificial intelligence (AI), metrics, and other data-driven techniques and technologies. It is the power that measurement, technical efficiency, and numbers hold over us, a power that gives an assurance of control, a conviction that [End Page xxxiii] there is an order to things, and a soothing sense of security. Our disquiet in the face of chance, uncertainty, and even tragedy is ultimately unjustified, for nothing truly out of the ordinary happens. There is always a larger pattern, if our technologies but give us the eyes to see it, that can be ascertained, measured, quantified, correlated, and technically mastered. At bottom, the world is a mathematical puzzle, and quantification is both the way to understand it and the means to solve it. The tremendous excitement for machine learning, predictive analytics, metrics, and their deployment in everything from smart cities to self-driving cars arises from their seeming novelty and the vast new possibilities they promise. Musil, however, whose book was first published in 1930, reminds us that the hopes and trust placed in mathematical reasoning are not so revolutionary, and long-ago mathematics "entered like a daemon into all aspects of our life" (Musil 1953, 40). These aspirations have a history. One of the larger themes of this issue, especially in the opening pieces by Joseph E. Davis and Daniel Doneson, is that what gives the new technologies of quantification their "force," what makes them compelling and moving, are ethical and philosophical commitments that have been unfolding since at least the dawn of the scientific revolution. The new technologies are an instantiation of these prior commitments—to control our fate, to possess an objective knowledge without a knower, to relieve the human condition—and their most singular expressions yet (and in the fantasy of "strong AI," their last). This history forms a critical part of the context in which quantification techniques and technologies are developed, deployed, and received. Across time, human rationality, once the standard of judgment and prudence, has undergone a fundamental revaluation; now "bias" is understood as human deviation from the rules of probability theory, statistics, and the computations of the machine. Ironically, arguments for data-centric technologies often are prefaced with the admonition to resist hubris and "beware of romanticizing human judgment." But who currently, in any of the fields that deal with cognition [End Page xxxiv] or neuroscience, is in danger of overestimating human judgment? Quite the opposite; it is the machine and the numbers that now hold epistemological superiority. The general climate is one of mistrust and loss of faith in human judgment, and in many contexts a more specific and forceful polemic is at work: in the chain of right reasoning, we are regarded as the weak, and ideally replaceable, link. And in the face of the onslaught of "big data," the feeble estimation of human judgment is becoming self-reinforcing. Familiar critiques of algorithms and AI are often insufficiently...