Material value ethics
The theory of material ethics of values is presented in Max Scheler’s Formalism in Ethics and the Material Ethics of Values, with the first part published in 1913 and the second in 1916 (Kelly 2007, 2011). Other phenomenologists such as Edmund Husserl and Nicolai Hartmann developed similar theories of material values during the early phenomenological movement. Schutz (1958) describes the premise of Scheler’s phenomenological material value-ethics (MVE) as follows: material or “… concrete values and their hierarchical order form a realm of material, aprioristic data which is disclosed to us by emotional intuition” (p. 486). Stated very simply, phenomenological MVE demonstrates that we can obtain a fairly detailed picture of what human being value if we abstract values from valued objects, as we might abstract the color blue from blue objects and focus upon the colors and their parameters themselves. The result will be a typology of values and disvalues and their relative worth. Values constitute a particular class of ideal objects, like numbers and figures that are objective and immutable. Accordingly, the emotional acts that intend them have epistemic value, and yet are beyond the grasp of the rational intellect alone. That fact does not suggest that our knowledge of values is random or chaotic, for feelings are not without an a priori order. Scheler quotes Pascal’s famous observation, “the heart has its reasons.” Indeed, it has been shown experimentally that the feelings of simple injustice in very small children are aligned with those of adults (cf. Bloom 2013; McAuliffe et al. 2020).
Furthermore, for Scheler material values are independent of our subjective bodily states and are intended by “pure” (that is, not visceral) emotional acts. That is why our knowledge of values remains the same even as our bodily states may vary chaotically from one moment to the next. For instance, our understanding of the phenomenon of sadness may remain unchanged, while our subjective or visceral emotional state changes from heartbroken to composed, and friendship will remain unchanged as a material value although we suffer when a friend betrays us. Values are independent of things and relationships of all kinds which are their carriers: the so-called “goods.” A value such as utility is similarly independent of our having something in use. Both can be thematized by phenomenology by repeating the intentional acts in which they are given. Thus, we can describe the material content of the value in question, just as Aristotle tries to describe in NE the content of the phenomenon of courage. By reflecting phenomenologically upon the level and intentionality of our emotions, it is possible to discover the realm of concrete values in an aprioristic way without deriving values from visceral feelings that the perception of empirical goods may cause in us. (cf. Scheler 2009, pp. 35–36).
No doubt for Scheler all mental awareness of self and world is made possible through a form of emotional affirmation and receptivity that is summarized for him in the word “love.” But values are given in specific emotional acts that re mediated by the structure of human sensibility. That structure is developed in Scheler’s Formalismus in der Ethik, but it would take us too far afield to present that structure in this paper. It is always operative in any attempt to exhibit the content of material values given in acts of feeling. Perhaps it suffices to point out that the kind and level of feelings that intend the value of courage in an act of self-sacrifice is quite different from the receptive emotionality of hearing a piece of music or the feelings directed at the various values revealed in the taste of a fine wine. Such values are given in emotional acts, as mathematical objects are given in rational intentional acts and do not exist apart from them. It is just because there is overlap in human sensibility that we can understand the values that function in cultures foreign to our own.
Scheler also outlined the aprioristic structure of the realm of values. First, all values are either positive or negative. Second, the whole realm of values is graded in an order of ranks in terms of where the values stand to one another in the relation of “higher” and “lower.” The gradation of rank is disclosed in the phenomenology of the pure emotions of “preferring” and “thinking less of.” Preferring refers to felt relationships among sets of values; it is an immediate feeling of the relationships of higher and lower prevailing among values. Scheler ascertains two different extant orders of rank of values. He first places values in accordance with their carriers; for example, personal values may have a higher rank than values for which goods are the carriers. The second order is the “modes” of values, where the lower value is founded upon a cognition of the higher one, which means the higher value is the axiological condition of the lower one. The order of rank of values, from the lowest to the highest, are: (1) the values revealed by sensory feelings, e.g., pleasure; (2) the values revealed by the class of vital feelings (utility), e.g., the feelings of health and sickness, courage, anxiety etc.; (3) the class of spiritual values (e.g., beauty, goodness); and (4) the values of the holy and the unholy (the sacred and the profane). We all recognize, for example, that collegiality and camaraderie are lower values than friendship—in communities, where these values function at all.
Based on this analysis, Scheler further argues that any ought-to-be (or ideal object) is founded upon some specific value or values. One ought (ideally, that is, in the absence of a specific case currently facing us) to be courageous and seek to save a drowning child, for he is a human being whose life possesses intrinsic value; I ought to be generous and give money to a beggar, for he is needy. Courage and generosity are called forth in such instances by the positive value of human life or the negative value of human neediness. We see then, how ideal entities, values, can become functional in human efforts to imagine and to achieve a world worth having. Values may be ideal objects and entirely independent of the real existence of their carriers. However, the ideal ought-to-be generates an obligation (ought-to-do), which refers to a potential volition aiming at the realization of the ideal value content. A person is a unique type of being who perceives intuitively these ideal a priori values that are carried upon possible objects and who also ranks the values as she acts within a situation. Moreover, the person is an absolute value, the concrete unity of intentional acts of different types and natures. Because personhood is present in each and every act, the acting person constitutes the whole of her actions and consequently can be morally accountable for them. Finally, values themselves have no power, as in Plato, to realize themselves in action. However, human beings sense themselves to be “called” to realize positive values possible in some situation. This call—an “ought to do”—does not emanate from a universal Kantian-type Categorical Imperative, nor from a duty to achieve a certain beneficial consequence, as in utilitarianism, but rather from the values themselves that the human agent perceives as the highest for him to realize in this situation. Of course, an agent’s behavior must be limited by moral rules: one should not commit murder regardless of the values realized from such an act. Scheler situates this experience of obligation within a process of individualization: laws must be flexible, for they are made for persons, not persons for laws. An example is given in Kelly (2011, p. 116) “[W]hen I experience emotionally the kindness of some action of a person towards some other person or other sentient creature, the moral value of kindness is given to me, and I respond to it in a specific act of affirmation. Similarly, once I grasp the validity of a demonstration of a theorem in mathematics, I naturally respond not only with intellectual assent, but also with a determination to use the theorem with confidence as a premise in further demonstrations”.
To grasp a material value is not the same as having those values function as an a priori within one’s own culture’s world- and value view in guiding its moral and other evaluative behaviors. We can understand the values functioning in an ancient Athenian’s patriotism (e.g., Pericles’ Funeral Oration in Thucydides), although those values may not function in the patriotism common today. We can understand what might have driven men a century or two ago to fight duels, though their values of honor and manliness, easily comprehensible to us, hardly function in the ethos of most cultures today. Of course the functionality of values evolves, although the essential content of the values themselves does not. Human cultures possess enormous diversity, though there is an internal structure of all value systems. In planning the alignment of AI and values by developing descriptions of key values we must always consider how they will function in machines, that is, how the material content of some values may function as an “a priori” to guide the “choices” among possible courses of action in contexts in which these robots will be put to use. It is important to bear in mind that the intentional acts that are aimed a phenomenon are only indirectly relevant to the project of this paper. It is the material content of any value described by phenomenologists that is of concern, not human re-enactments of the emotional acts in which they are given. As yet there is no equivalent of intentionality in AI.
Finally, Kelly (2011) claims that material values, fundamental to phenomenological axiology, bring both concrete and synthetic understanding to human values. It offers a systematic means towards a personal response to the Socratic question: how should we live? “There are many incompatible ways of living successfully and happily, but they are all (should be) founded in the right knowledge of the values themselves” (Authors’ emphasis). Consequently, any moral agent must be capable of understanding the nature of some set of values, of perceiving values as “carried upon” objects, processes or actions, and of having the means for realizing specific valued objects and processes. Human beings are also able to relate their actions to their own specific history and to have the flexibility to order their actions with reference to that personal history. That is the foundation of human integrity.
Technological mediation theory and its significance for AI
We offer here another analysis of the interface between human being and the technology that gives us insight to our present efforts to explore the question of AI and values. This theory was advanced by Don Ihde (1990), a critic within the aegis of phenomenology. In criticism of Heidegger’s (1954) account of technology, he argues that his theory is too abstract and alienates technology from human use. Heidegger (1954) is not aiming at practical use but at the meaning of being that holds sway in eras characterized by the dominance of technology and that disrupts our ability to let things appear as the things they are. Heidegger’s (1954) theory does not pay sufficient attention to the actual experiences people have of the roles of technologies in human existence. To address this concern, he develops the technological mediation theory, which demonstrates how technology mediates human experiences and perceptions with the lifeworld. Technology is analyzed in terms of the relations between human beings and technological artifacts with the focus on interpreting the different ways that technologies shape relations between human beings and their world (environment). It regards technologies as the mediators of human experiences and practices rather than merely as functional and instrumental “objects”.
Ihde (1990) distinguished in his analysis four types of relations between technology and human beings. First, technologies can be embodied by the users (embodiment relation), such as the glasses worn to see better. Second, they can be the terminus of our experience (a hermeneutic relation), for example, we can buy a bus ticket from a ticket machine. Third, technologies can give a representation of reality (an alterity relation), for instance, a thermometer measures a number of temperatures without producing the reality of heat or cold; and fourth, technologies can play a role at the background of our experience, creating a context for our perceptions, such as public video surveillance systems installed in many big cities.
Ihde calls his approach to values “post-phenomenological.” His concept of multi-stability is relevant here. For, he notes, “no technology is ‘one thing,’ nor is it incapable of belonging to multiple contexts” (1999, p. 47), that is, the same technology can have multiple instantiations in history or across cultures, each of which may be stable in each instance. Multi-stability also means that a technology can be put to multiple purposes within multiple constellations of values and thus be relevant and useful in different ways to different users. The concept of multi-stability in human–technology relations functions within multiple embodiment or hermeneutic relations in a given human praxis. It is remarkable how the living systems functioning in organisms have been altered and adapted by the evolutionary process to function in new ways in different organisms at different temporal points. Technology that was developed for specific purposes in the functioning of routines may similarly be repurposed as the complexity of AI grows.
Given that fact, this multi-stability of technologies makes it nearly impossible for designers to anticipate the ways in which given technologies will influence human actions and then to evaluate this influence in a system of values. Who could have predicted, in 1903, to what uses the Wright brothers’ invention would be put and how it would transform our lifeworld and even determine the values that function a priori in our consciousness of ourselves in that world? Because of the multi-stability factor, designers are not able to maintain an equivocal relationship between their activities and the mediating role of the technologies they are designing; moreover, the technological mediations emerge in a complex interplay between technologies and their users. Technologies have no fixed identity, for they are defined in the context of their use and are always “interpreted” and “appropriated” by their users. Verbeek (2011) describes in general how the forms of agency that appear through technologically mediated human actions may be interpreted. There is “(1) the agency of the human being performing the action or making the decision to do so in interaction with the technology and appropriating the technological artefact in a specific way; (2) the agency of the designer, who, either unintentionally or [deliberately], give a shape to the technology and thus helps to shape its eventual mediating roles; and (3) the agency of the technology [that] mediate[s] human actions and decisions, sometimes in unforeseen ways” (p. 99). To handle the complexity of technological mediation, designers should make a connection between the context of design and the context of use with the aim not only to formulate technical features, such as technical artefacts, affordances, and symbolic expressions (e.g., Markus and Silver 2008), but also to obtain at least an informed prediction of the technology’s future mediating roles. Consequently, the role of a material value ethics in guiding the alignment of AI with human values must be as flexible and relative to circumstances as are human values themselves, which may become functional in new ways as conditions evolve and become subject to unexpected injections of new values in a non-closed system of values.
At this point, we may summarize. Technological mediation theory argues that technologies and humans co-constitute the context, feelings and experiences as humans design and use technology. People “feel” the world about them in new ways and thus discover new values or new functions for those that are already recognized among them. Thus, this new material content of values in our lifeworld may generate different “feelings” and experiences of values for different persons, as well the dynamic development of knowledge of values in the process. In this reasoning, we think that Ihde’s work paves the way for us to understand the material content of values that are discovered and functionalized by technology and humans in tandem, as humans design technology and use it to (act upon) the world and as the new world contexts they have created act back upon them. AI must situate itself within this process so that we may align it with our developing experience of values.