When the Machine Takes The Wheel – Towards a Posthumanist Understanding of the Moral Agency of Waymo’s Autonomous Vehicles
Student name: Sara Vuorio Student number: 11316470 Program: rMA Cultural Analysis
Supervisor: Dr. Daan Wesselman Reader: Niall Martin
CONTENTS
INTRODUCTION...3
CHAPTER 1: THE VIRTUAL-MATERIAL DRIVER-CAR...11
ENVELOPED RULES AND VALUES...15
PERCEPTION AND INTENTION BY PROXY...18
MORAL AGENCY IN NETWORKS...25
CHAPTER 2: IMAGINARY AFFORDANCES AND AFFECTIVE AESTHETICS...30
FROM IMAGINED AFFORDANCES TO MATERIAL CONFIGURATIONS...32
POSTHUMAN AESTHETICS AND DESIGN...36
AFFECTIVE ENCOUNTERS...41
CHAPTER 3: THE URBAN ASSEMBLAGE...45
THE CYBORG CITY...48
THE URBAN ASSEMBLAGE...53
DISTRIBUTED AND SPATIALIZED MORALITY...57
CONCLUSION...60
WORKS CITED...63
INTRODUCTION
The autonomous vehicle is often seen as a currently developing technology that aims to improve mobility and road safety. Since the 1930s, the motor car has been a global key
commodity in production and consumption that “shapes the built environment, cuts through the landscape and dominates the soundscape” (Dant 61). The motor car, that Henri Lefebvre once – in terms of its centrality within the culture of modern societies – referred to as ‘the Leading-Object’ (Lefebvre 100), is being currently being transformed by self-driving technology. According to current estimates, autonomous vehicles are likely to become the primary means of transport by 2050.1 While levels of uncertainty currently surround the impact of such technology, the accelerating development of autonomous vehicles enclose a significant promise for their potential to transform mobility, economy and the society as a whole.
An autonomous vehicle operates on an AI system that drives the vehicle via localization, mapping, scene perception, vehicle control, trajectory optimization and high-level decision planning. The self-driving technology utilizes machine learning algorithms that are based on human-annotated data sets. Self-driving technology’s Machine Learning applications include evaluation of driver condition or driving scenario classification through data fusion from different external and internal sensors.2 Until the most recent developments, the AI system in self-driving technology has been able to perform anything from just over 0% to just under 100% of the driving.3 As such, autonomous vehicles represent so called cognitive technologies, that imply a trajectory towards greater technological agency and autonomy (Hayles 45). Taking up the driving task, self-driving technology enables a form of social action that has previously been considered to be beyond technological abilities. On the other hand, the underlying intimation of self-driving technology is the idea of the unreliability of the human driver. According to the World Health Organization’s statistics, more than 1.25
1 For more detail, see: https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/ten-ways-autonomous-driving-could-redefine-the-automotive-world
2 For full detail, see: https://docs.google.com/document/d/1HIuTNmZvXnVh1oiwfsZJYrDSsekqAK-XumK1c8D4g7Q/edit
million people die each year as a result of road traffic crashes. On a national level, in the U.S. 94% of the crashes involve human choice or error, including speeding, intoxication, distraction or drowsiness.4 Thus, at the current stage of technological development, removing the human input from the driving task has been considered as a solution for safer, easier mobility.
In November 2017, Waymo, a company that started out eight years earlier as Google’s self-driving technology project, announced that their new Level Four self-driving vehicles would start test-drives on public roads without a human driver in the driver’s seat.5 The new model of Waymo vehicles uses sensors that detect its surroundings, and is equipped with safety features that guarantee ‘full autonomy’ for the vehicle.6 Prior to the launch, in February 2016, the American National Highway Traffic Safety Administration (NHTSA) stated in a letter to Google’s Director of the Self-Driving Car Project in reference to their fully autonomous vehicle design that:
NHTSA will interpret ‘driver’ in the context of Google’s described motor vehicle design as referring to the SDS [self-driving-system], and not to any of the vehicle occupants […] even if it were possible for a human occupant to determine the location of Google’s steering control system, and sit ‘immediately behind’ it, that human occupant would not be capable of actually driving the vehicle as described by Google. If no human occupant of the vehicle can actually drive the vehicle, it is more reasonable to identify the ‘driver’ as whatever (as
4 For more detail, see: http://www.who.int/mediacentre/factsheets/fs358/en/ and https://waymo.com/tech/
5 For more detail, see here: https://medium.com/waymo/with-waymo-in-the-drivers-seat-fully-self-driving-vehicles-can-transform-the-way-we-get-around-75e9622e829a
6 The levels of driving automation of autonomous vehicles are designated by Society of Automotive Engineers (SAE). Level Four indicates that the vehicle cannot be driven by the humans inside it, rendering the AI system as the sole driver of the vehicle. For a more detailed account, see: https://www.itf-oecd.org/sites/default/files/docs/15cpb_autonomousdriving.pdf
opposed to whoever) is doing the driving. In this instance, an item of motor vehicle equipment, the SDS, is actually driving the vehicle7
As evident here, NHTSA identifies the technological system as the driver of an autonomous vehicle. Public policy and statements such as this have encouraged the development of self-driving technology. Since the launch of their Level Four vehicle, Waymo has been working towards building a fleet of fully autonomous vehicles that drive on public roads. In addition, the vehicles are being tested on closed test courses and in Waymo’s simulation software. The company’s aim is to produce vehicles for driverless ride-hailing service, logistics, to public transport and personal vehicles.
The function, use, social and economic value of the automobile has developed hand in hand with the society, where, “the relevant groups have […] invested so much in the artefact that its meaning has become quite fixed.” (Bijker 282). As Robert Kirkman argues, this has led to an establishment of a “technological frame” that has served to constrain and determine the production of sociality and space (Kirkman 237). The urban environment has become what John Urry calls an “automobilised time-space”: The spatio-temporalities of whole cities have been designed and planned, built and lit to accommodate the automobile (Urry 59). This automobilised time-space consists of streets and sidewalks, bridges, tunnels, highways, cars, trucks, buses, powerlines, traffic signals and many more components, supported and enabled by industrial and infrastructural production. Furthermore, the figure as well as the social actions performed by the automobile have also become fixed and habitual within the society. As Tim Dant has outlined, the driver-car assemblage8 is a social being that produces a wide range of social actions including “driving, transporting, parking, consuming, polluting, killing and communicating” (Dant 62). For instance, driving, the key social function of the driver-car 7 For full detail, see:https://isearch.nhtsa.gov/files/Google%20--%20compiled%20response%20to %2012%20Nov%20%2015%20interp%20request%20--%204%20Feb%2016%20final.htm
8 I borrow this term from Tim Dant so as to refer to the collaboration of human and the machine that merges their subjectivities to create a new form of social being.
has been undoubtedly treated as something a human being does to the car, inside the car, on the road (Dant 54). The car, on the other hand, has been treated as a ‘locomotion tool’, affording the human driver–the agent of any ensuing action–mobility (Gibson 1082; Dant 63).
As I see it, as the first fully autonomous vehicle, Waymo produces a disruption to the common understanding of the driver-car assemblage. Indeed, as Nigel Thrift argues, the transformation of cars are producing a new kind of phenomenology that relies on the concepts of memories, intelligence, networks, interfaces, feedback loops and software (Thrift 49). Waymo’s AI system controls the vehicle’s functioning from engine management to brakes, lights, speed, parking manoeuvres, communication, navigation and crash protection systems and more; the whole technical ecology of driving. This marks for a redistribution of agency within the driver-car assemblage, a rearrangement of the collaborative relationship between human and the machine that reaches beyond the vehicle itself.
Autonomous vehicles are by no means the first technology that has been considered to suggest a redistribution of agency via redistribution of cognition. The envelopment of ever-smarter technologies that perform routine and non-routine tasks for humans has become so pervasive that technological agency has produced a “new technological unconscious of a world of performative infrastructures” (Thrift 197). What makes self-driving technology stand out within this process is the emphasized moral dimensions of its agency. Indeed, whilst the introduction of autonomous vehicles has stimulated a large public debate on various aspects of their social, cultural and economic impacts, one of the most discussed features has been their moral agency - or their lack thereof. Illustrating a common stance on autonomous vehicles’ inappropriateness for the consideration of moral agency, tech journalist David Chang writes: “Because autonomous vehicles lack both conscious intentionality and responsibility, they cannot have moral agency. It is therefore more appropriate to attribute
any ethical decisions to the manufacturers, rather than to the robots themselves.” (Chang 2017). Chang’s claim derives from the opposition that believes that computer systems do not meet the standards we hold for moral agency, and thus cannot and should not be considered as having moral agency.9 Thus, as I see it, it is the moral aspects of the agency of Waymo that make their introduction disruptive within the driver-car assemblage.
The idea of moral agency and the very idea of morality have been discussed throughout the history of Western moral philosophy. Historically the debate has resurfaced at the events of the introductions of new technologies, often as a result of general anxiety over the moral implications of technologies (Mokyr, Vickers & Ziebarth 32). As evident here, removing direct human input from driving as well as classifying the AI system operating the vehicle as the driver raises complexities in relation to our understanding of moral agency and moral subjecthood. The base of the common understanding of moral agency relies on Kant’s formulation of the concept as the individual’s ability to act based on internal notions of right and wrong, and be held accountable for these actions. Conceptually, moral agency has been separated from the general idea of agency because the ability to act does not presume a moral dimension to the capacity to make a decision to, i.e. act morally. From this perspective, it is legitimate to argue, as asserted by Deborah Johnson, that since computer systems do not have mental states, they cannot and should not be considered moral agents (Johnson 195). Borrowing from the Cartesian epistemology, the claim that computer systems have no moral agency relies on the dualisms between human/nonhuman, mind/body, subject/object and natural/human-made. However, as Bruno Latour has famously argued, these dichotomies are no longer valid, or, indeed, have never been valid. Instead, we live in a collective entanglement of humans and nonhumans. Moreover, this pervasive enmeshment has led to a point where we can no longer meaningfully separate the human from the nonhuman.
9 Deborah Johnson provides an excellent example of this account,: see: Computer systems: Moral entities but not moral agents. (2006).
As I see it, the times we live in force a renegotiation of the locus and nature of moral agency. The rapid development of advanced technologies has led to a point where artificial intelligence is implemented in technologies that have autonomy to perform tasks that illustrate moral agency. Therefore, there appears to be a lack of balance between our technological development and its conceptual, ethical and cultural roots (Floridi 154). This imbalance, according to Luciano Floridi, derives from philosophy’s inability to deal with nuances has resulted in “mass distraction” from the pressing moral issues concerning artificial intelligence (Floridi 2017). Fixed notions of moral agency prevent the building of solid foundations for technological agency and for examining how computer systems function as part of morality.
The enmeshment of humans and nonhumans has led to the rise of posthumanism within academic discourse. The notion of posthumanism correlates with the postmodernist and poststructuralist currents of thought, while devoting much-needed attention to nonhuman others. Rejecting the dichotomy, posthumanist theory aims to think about humans and nonhumans as “a non-hierarchical collective” (Radomska 94). Philosophy’s dismissal of machines from the realm of moral agency is built on anthropocentric, humanist premises that have already been called to question by the posthumanist theory on many accounts. However, the debate around moral agency still draws largely upon our philosophically-rooted humanist conceptualization of moral agency that disregards the fact that humanity is a part of a complex system and thus cannot be treated as if it exists in isolation.
Monolithic concepts such as moral agency can be broken apart and re-examined by thinking through the lens of agential realism, a posthumanist ontology introduced by physicist Karen Barad. Barad’s ontological formulation is based on an integral refusal of anthropocentrism as well as a rejection of representationalism, both of which have dominated Western thought up until now. For Barad, representationalism “separates the world into
ontologically disjunct domains of words and things, leaving itself with the dilemma of their linkage such that knowledge is possible” (Barad 137). Posthumanism, on the other hand, does not “presume separateness of any pre-existing entities” (Barad 136). From this perspective, agency does not pre-exist separately, but instead emerges from the material relationships in intra-actions. What is significant here is that for Barad, matter is always agentive, “always about differentiating and mattering” (Radomska 92). For Barad, the boundaries, properties and components of every phenomena, as well as their meaning, are constituted through intra-actions. Drawing upon Michel Foucault’s notion of discursive practices10, and Niels Bohr’s concept of apparatus11, Barad formulates material-discursive practices as the primary semantic units (Barad 141).
Barad’s performative understanding contests the metaphysical assumption that agency is an inherent attribute, existing independently of a being’s intra-actions. It acknowledges the dynamic reality where moral agency is not an individual quality but a constant reconfiguration of the world (Barad 137). By assuming a performative approach, we can go beyond fixed descriptions and notions of agency, and shift the focus onto the matters of practice, action, doing and participation from which agency emerges. This represents a Deleuzoguattarian approach to philosophical concepts as “centers of vibrations”; “configurations, constellations of the events to come” (Deleuze & Guattari 23).
A posthumanist performative approach does not consider the concepts of agency and moral agency as disjointed, but thinks of them as being together, intertwined, entangled. Barad’s understanding of posthumanism describes an ethico-onto-epistemological structure, where matter and meaning are always already entangled with ethics (Radomska 110). As 10 For Foucault discursive practices are “the sociohistorical material conditions”, which enable the practices of disciplinary knowledge. These conditions are always immanent and historically situated. 11 The Bohrian concept of the apparatuses is the “particular, physical arrangements that give meaning to certain concepts to the exclusion of others; [...] the local physical conditions that enable and constrain knowledge practices such as conceptualising and measuring; they are productive of (and part of) the phenomena produced; they enact a local cut that produces “objects” of particular knowledge”.
such, as I see it, it provides a useful contemporary framework to examine AI systems that are implemented in technologies that carry out tasks with moral significance, such as Waymo. Thus it is the intention of this thesis to bring Waymo into a discussion with the emerging scholarship on the posthumanist performativity that speaks to the issues of agency and entanglement. The idea is not to discuss the role of moral agency to describe Waymo as an implicit, explicit or full moral agent, but rather to use it to examine how Waymo functions as a part of our moral system. As the first fully autonomous self-driving vehicle driving on public roads, Waymo challenges our understanding of moral agency and the moral subject and subverts the humanist dualisms between human/nonhuman, mind/body and technology/nature. Thinking about Waymo through the lens of intra-action allows a conceptualization of Waymo as a material-discursive, socio-technological system that has a significant impact upon the production of urban space.
To begin, I aim to characterize moral agency through different models of agentic functioning that foreground Barad’s conceptualization of intra-active agency in relation to Waymo. Through calling into qnuestion the anthropocentric confinements of the concept of moral agency via a posthumanist performative approach, this chapter argues that the moral agency of Waymo emerges from the intra-actions within the human-nonhuman relationship. Moving further, this chapter will then consider how Waymo’s functioning as a driver-car assemblage conditioned by complex virtual-material networks. Considering Waymo as a disruption within the driver-car assemblage, this chapter argues that Waymo and the nonhuman agencies inherent to its networks produce a new kind of networked moral subject that undermines human exceptionalism and the singularity of human morality. The second chapter will then extend this discussion through analyzing how Waymo’s moral agency is tied to aesthetics and affect. Through employing the concept of imagined affordance informed by the work of Peter Nagy and Gina Neff, I will argue here that Waymo’s moral
agency mediates through users’ perceptions, attitudes and expectations that are conditioned by the materiality and functionality of Waymo, as well as its aesthetic design. In the third chapter, I will discuss Waymo in relation to the urban environment. Here, I will build by analysis on the concept of the assemblage as understood by critical urban theory to examine how Waymo’s moral agency is entangled with the production of urban space.
CHAPTER 1: THE VIRTUAL-MATERIAL DRIVER-CAR
In an online video that can be found on the Waymo website, viewers are introduced the technology that powers Waymo’s autonomous vehicles, as well as what it is like to ride with Waymo. The video portrays several key features of Waymo’s technology, including its navigation and sensor systems. Documenting a typical ride with Waymo, the video shows how Waymo’s technology functions on the road in traffic. For instance, the video shows Waymo’s cameras detecting a red traffic signal and stopping at a crossroads. A female voice-over explains how at that “fraction of a second” Waymo combines all the data it receives from its current environment and its AI system to “understand the world around it”; that is,
to identify surrounding objects, predict their behavior and plan the vehicle’s actions accordingly. A moment later, the video shows Waymo giving way to a cyclist “while also looking out for” a pedestrian on a nearby sidewalk. This, according to the voice-over, is a good example of how Waymo’s decision-making in planning a safe path is guided by accounts for both the passengers’ and the surrounding people’s safety. (Youtube Waymo 360 Experience). It is evident from the video that Waymo sees, interprets and acts in the material world according to perceptual and cognitive systems that are embedded in its technology. Moreover, what is striking to the viewer is how Waymo functions and interacts with other objects in a constant reconfiguration of the world. And as it does so, it combines and transforms bodies into data, extending them into complex systems and networks.
As I see it, this process of embodiment can be reflected upon through what Stacy Alaimo has called ‘trans-corporeality’, which describes the imbrication of material human bodies not only with each other, but also with nonhuman bodies and material environments (Alaimo 18). From a trans-corporeal perspective, concepts that have long been considered exclusively ‘human’, such as agency, are dispersed onto “a wider, and messier, field of matter” (Alaimo 28). Referencing Moira Gatens’ reading of Spinoza, Alaimo argues that since the body is always in constant interchange with its environment, it is always “radically open to its surroundings and can be composed, recomposed and decomposed by other bodies” (Gatens 110; Alaimo 24). In its perpetual reconfiguration of the world, Waymo expresses this sense of intercorporeality. As shown in the video, for instance, the material body of the cyclist transforms into a virtual body in Waymo’s system that becomes a part of software data log which extends into a web of simulations that, in turn, condition how Waymo drives on real-life public roads. This forms a looped assemblage where material and virtual bodies become constituted and reconstituted in relation to one another. As Katherine Hayles argues, this sort of construction is what forms the posthuman body, “for the enacted and represented
bodies are brought into conjunction through the technology that connects them” (Hayles How xiii). This, as I see it, is why a posthumanist account can serve to reflect on these critical issues, as it is known to pay special attention to the material and corporeal dimensions in the formation of subjects and subjecthood.
The video evokes a consideration of Waymo as a technology connecting bodies, virtual and material practices and systems of knowledge in an ongoing reconfiguration. As such, it alludes to the idea of a materially contingent but endlessly malleable assemblage, that results from the collaboration of human and machine: the driver-car, a “socially embedded system of affordances, actor networks and embodiment” (Dant 75). As I see it, Waymo marks the transformation of the traditional driver-car assemblage into a social being that introduces new forms of social actions and human-machine relations. The driver-car assemblage of Waymo becomes constituted through the entanglement of software and material bodies and environment. As such, it is a product of human design, manufacture and choice (Dant 62). Unlike the traditional driver-car assemblage, however, Waymo is not dependent on the human driver to afford mobility. Through its software, Waymo extends the driver-car assemblage to virtual worlds, algorithms and the people who wrote them. As Katherine Hayles argues, humans and technical systems form an interconnected assemblage, where cognition is distributed within the system (Hayles Cognitive 33). Thus the driver-car assemblage of Waymo suggests more than a collaboration between the human and the machine - ‘the cyborg’ - but an assemblage that encompasses an array of virtual and material spaces and bodies.
As Margit Shildrick remarks, the notion of assemblage raises not only ontological and epistemological, but also ethical reconfigurations (Shildrick 24). In this sense, matter comes to matter through its entanglement with ethics. Alaimo characterizes this entanglement through Gail Weiss’ conceptualization of the ethics of intercorporeality as a continual
interaction with other human and nonhuman bodies, where ethical behavior is provoked by “calling us to respond ethically to one another” (Weiss 5; Alaimo 23). Alaimo’s stance corresponds with Barad’s agential realism, where “ontology is always already intertwined with epistemology and ethics” (Radomska 102). Her scholarship can be described as an ‘ethico-onto-epistemology’ that accounts for the dynamic production of matter and meaning, “both always already entwined with the issues of responsibility and accountability for the relations in their ongoing reconfigurings” (Radomska 110). Thus, Barad proposes that the questions of ontology, materiality and agency have to be approached with a posthumanist performative understanding, leading to “understanding technoscientific and other naturalcultural practices that specifically acknowledges and takes account of matter’s dynamism” (Barad, Meeting 135). Waymo’s continual reconfiguration of the world resonates with Barad’s notion of intra-action, where “the boundaries and properties of the components of phenomena are being constituted” (Radomska 104). Drawing upon this, Waymo becomes conditioned by the entanglement of intra-acting agencies. As such, the posthumanist performative understanding advocates the emergence of alternative, non-representational ways of thinking about moral agency.
Thus, the intention of this chapter is to bring Waymo into a discussion with the emerging scholarship of posthumanist performativity that speaks with issues of agency and material-discursive practices. To begin, I will address how discourse conditions the moral values enveloped in Waymo’s technology, before examining the technological affordances that enable Waymo to function as part of the moral system. Moving further, I will contemplate Waymo’s agentic functioning through the notions of perception and intention in relation to the concept of proxy agency. Within the framework of Bruno Latour’s Actor-Network Theory, I will then move on to explore how Waymo’s driver-car assemblage is conditioned by complex virtual-material networks. This chapter argues that the moral agency
of Waymo emerges from the intra-actions within and between material-virtual bodies and spaces via perception and experience. Considering the networked condition of the driver-car assemblage encompassing human and nonhuman agencies, this chapter asserts that Waymo produces a new kind of networked moral subject.
ENVELOPED RULES AND VALUES
As social studies of technology have shown, the design of technology is conditioned by social and cultural norms and rules, which means technology is never neutral from politics or values (Nagy & Neff, Talking 4922). Furthermore, technology does not only reflect our sociocultural values but also has the potential to influence them. As Katherine Hayles argues, human decisions and interpretations feed back into technological systems, affecting the contexts in which the systems operate (Hayles Cognitive 33). In terms of identifying the values that factor into Waymo, the company website offers an entry point for analysis. As described on the website, Waymo’s mission is to “make it safe and easy for everyone to get around”. Moreover, the website evokes the visitor to “imagine if everyone could get around easily and safely, without tired, drunk or distracted driving”. (Waymo.com). In addition, Waymo’s goal is outlined to be “working to make our roads safer, free up people’s time, and improve mobility to everyone” (Waymo.com/Journey). The systematic nature of vehicle crashes is something that has been denied by the automobility industry. Through framing the human driver as being potentially morally irresponsible, i.e. conceivably drunk, tired or distracted, the website presents Waymo as the anomaly that deviates from this standard. To answer the question of why Waymo is working on fully self-driving cars, the Waymo website states:
Annually, over 1.2 million people die on our roadways. In the US alone, traffic collisions kill over 35,000 people a year–the equivalent of a 737 airplane falling out of the sky every single working day of the year– and that number is rising. In the U.S., 94% of crashes involve
human error or choice, and this is one place where we believe we really can bring technology to bear. (Waymo.com/faq).
This implied juncture between ‘human error or choice’ and car crashes highlights the fallibility of the morality of the human driver. Furthermore, it promises a lower, more tolerable level of traffic accidents in the case of Waymo. Thus, human safety appears to be a core value and a mindset for Waymo. Waymo’s understanding of safety mirrors anthropocentric confinements of the safety paradigm. The lack of reflection on the safety of the vehicle, for example, discursively draws attention away from the idea of car crashes. “Every day, our cars drive safely through many complex scenarios on real city streets (Waymo.com/faq). Furthermore, the company’s broader goals imply values of freedom, accessibility and equality. For example, one of the company’s claims is to aim to improve mobility for elderly and people with disabilities. As stated on the website, Waymo is committed to helping people who can’t drive “to get around and do the things they love” (Waymo.com/faq). .
What is specific to Waymo is how these values are encoded into Waymo’s technology and how Waymo produces modes of agentic functioning through its technology and its systems of knowledge. That is, as stated on the Waymo website, “a range of technologies we’ve built from the ground up to understand the world around [Waymo] and get you where you need to go” (Waymo.com/Tech). Moreover, as the director of Stanford University’s Center for Automotive Research Chris Gerdes describes:
It [Waymo] will understand our rules, communicate its desires, be legible to our eyes and minds. […] They [Waymo] understand the world. And then for anything that is a dynamic actor in the environment—a car, a pedestrian, a cyclist, a motorcycle—our cars understand intent. It’s not enough to just track a thing through a space. You have to understand what it is doing. (The Atlantic)
As I see it, Gerdes’s statement precisely illustrates the Baradian idea of the intertwined status of ontology, epistemology and ethics, where both the material and the discursive have a role. Furthermore, Gerdes’s remarks highlight the importance of understanding the conceptual nature of behavior in human-machine communication. As argued by Heather Love, behavior can be considered the basis for the “production of meaning, intention and culture” (Love 412). The agentic behavior of Waymo is conditioned by the behavior of other human and nonhuman bodies and how that is enveloped in algorithms and transformed back to the material vehicles. As such, Waymo becomes a material component of algorithmic modeling. As Waymo’s vice president of engineering, Dmitri Dolgov explains, “[The] sort of understanding of the behaviors of other participants in the world, is very similar to this task of modeling them in simulation.” These remarks indicate an understanding of rules, intent and behavior as discursive; something that can be algorithmically modeled. Furthermore, it corresponds with a performative understanding of agency as a matter of practice, doings and actions.
For context, I would briefly like to give an example of how this interplay has been approached in terms of remapping moral behavior through algorithmic modeling. A recent study conducted at the University of Osnabrück studied human behavior and moral assessments through simulated road traffic scenarios in virtual reality. The results of the study suggest that human morals and behavior can be algorithmically modeled by a single value-of-life for human and nonhuman objects. The first author of the study, Leon Suetfeld, argues that “Human moral decision can in principle be explained by rules, and these rules can be adopted by machines.”12 As suggested here, the moral system can be broken apart into rules
12 For more detail, see:
https://www.uni- osnabrueck.de/presse_oeffentlichkeit/presseportal/pressemeldung/artikel/machines-will-soon-be-able-to-imitate-human-moral-behavior-research-results-from-the-institute-of.html
and assessments that can be algorithmically modeled, which subverts both the singularity of human morality and the assumption that moral decisions are strongly context-dependent.
However, following Barad’s line of thought, it can be argued that agency is only realized through multiple practices that are material as well as discursive. As Luciano Floridi asserts, within the information society the ontological focus has shifted from a materialist one to a dephysicalized, typified understanding of concepts (Floridi 156). This, however, overlooks the material dimension of the cultural or discursive constrains. According to Barad, it is only at the specific material reconfigurings where “matter comes to matter” and meanings and boundaries of things are constituted (Barad Meeting 140). As I see it, the concept of behavior resonates with Barad’s remarks. For instance, Waymo’s behavior consists of a range of actions that are conditioned by the moral rules that are enveloped in its technology. The meanings of these actions emerge in conjunction with the material environment of Waymo and the intra-actions with other systems and organisms: humans, cars, traffic lights, curbs, lanes, intersections and roundabouts. This illustrates the entanglement of Waymo’s virtual and material affordances at the level of the specific material arrangement of the vehicle. The ways in which Waymo “sees” the world are tied to the ways in which Waymo “understands” the world, which shows how perception foregrounds Waymo’s agentic functioning. As Katherine Hayles argues, new forms of ‘distributed cognition’ are creating an integrated system of sensory perception that challenge the humanist understanding of consciousness and experience (Hayles How 228). To consider this point further, I now wish to turn to explore this entanglement through discussing the questions of perception and intention via the concept of proxy agency.
PERCEPTION AND INTENTION BY PROXY
As I see it, what enables Waymo to function as part of morality is the networked condition of its technology, which implies a redistribution of moral agency via distribution of the shared set of moral rules and values embedded in Waymo through its algorithmic modeling. This enables Waymo to function as part of the moral system. The set of moral values embedded into its algorithms conditions how Waymo interprets, makes decisions and acts. Take, for example, the previously discussed moment in the online video about Waymo’s technology whereby Waymo gives way for the cyclist it identifies across the road (Figure 1). In the video we see Waymo adjusting its speed according to the speed and the vehicle’s distance from the cyclist (pink), while maintaining a safe distance from other vehicles (blue). The animated trajectory (green) illustrates the “the safe path ahead” that Waymo’s AI system plans by transforming material cues into virtual data. In an effort to articulate an understanding of Waymo’s moral agency here, I wish to discuss the notion of affordance, which refers to the way in which animate beings relate to their material environment . Introduced by James Gibson, the concept of affordance refers to properties a particular material species affords. As such, it does not refer to general, fixed features but relational qualities. Gibson describes affordance as a process whereby a certain species “fits in” with some material aspect of its setting; a complementarity of the species and environment. (Dant 64). The strength of the concept lies in the ways it establishes the material relations between objects and their environment. As such, it can be used to describe human-object relationships within the society, including human-technology relations.
Describing the affordances related to the traditional motor car, Tim Dant writes: “The motor car affords the human being locomotion and mobility, and it affords the driver motility (the capacity to move spontaneously and independently) (Dant 65). Dant’s characterization of the motor car goes towards explaining the human intentionality that creates the affordances of the car. For example, the above described situation at an intersection where Waymo gives way to the cyclist reflects human intentionality that materializes through the vehicle’s behavior in traffic. As such, Waymo illustrates affordance theory’s claim that technologies produce fields of actions, “but that not all actions are possible” (Neff et. al. 202). In relation to Waymo’s moral agency, the “not all actions” indicates acts that would be, for example, to put it in Bandura’s words, “inhumane” actions, e.g. hurting the cyclist instead of giving him room to cycle past the vehicle. As is made evident here, the field of actions that Waymo’s affordances produce is conditioned by the rules and norms coded into its technology.
However, as he further argues, the car cannot simply be treated as offering affordance to its driver, as that obscures the complex process from which the driver-car assemblage
emerges (Dant 74). Furthermore, According to Dant, driver-car takes on properties of both the human and the machine, producing a social being that cannot exist without both. Waymo,
20
however, reshapes the relations of affordances between humans and cars. For instance, Waymo can afford mobility without any humans inside the vehicle and thus is not dependent on the driver’s motility. Furthermore, Waymo challenges the concept’s presumption of a prior physical relationship between things. Combining virtual and material worlds and bodies, Waymo subverts the idea of representational, separate entities and prior relationships between objects. Within the virtual-material driver car of Waymo, affordances become entangled, intertwined and multi-relational.
The relationship between Waymo’s technological affordances and human intentionality can be further considered in contrast to Albert Bandura’s concept of proxy agency, which explores human agency via social cognitive theory. For instance, Bandura distinguishes a socially mediated form of human agency termed ‘proxy agency’, that people exercise in order to have others act on their behalf to receive the outcomes they themselves desire (Bandura Exercise 75). From the perspective of technology, proxy agency can be considered through the intra-actions between humans and technological artifacts. As Nagy and Neff argue, when people interact with technology, they create a proxy agentic relationship through a technologically mediated entanglement of agencies (Nagy & Neff, Talking 4926) This definition of technological proxy agency corresponds with the previously discussed motivations behind the development of self-driving technology and the desire to delegate the driving task to machines. For instance, the form through which riding with Waymo is presented here describes the relationship between human intentionality and the desired outcomes of Waymo:
“Some things might stand out: There is no one on the driver’s seat, no one turning the wheel […] You might also notice that the drive feels a lot like driving in a regular car. And that’s the way it should feel. All this technology that allows Waymo to see, identify, predict and plan should make an extraordinary ride to feel completely
ordinary and create a very normal journey that simply gets you from A to B, safe and sound. All you have to do is sit back, relax, and enjoy the ride” (Youtube Waymo 360 Experience)
As is made evident here, the notion of proxy agency emphasizes the outcomes of actions. It is implied here that the forms of action performed does not change significantly within a proxy agentic human-technology relationship. The quote prompts the idea that that the agentic properties that are delegated to Waymo enable it to perform the driving task to create “a very normal journey”. Furthermore, it evokes the question: What constitutes driving “normally”, and not just, for instance, “legally”? As such, the quote emphasizes the social dimension of driving. The agentic functioning between humans and Waymo show how humans attribute agency to technological artifacts, resembling how Bandura’s proxy agency is practiced among humans.
Furthermore, inherent to Bandura’s concept of proxy agency is the idea of a tension between desire and lack. That is, people exert their influence on other beings out of a desire to receive an outcome that they themselves lack the means to produce. In relation to Waymo, this agentic functioning can be considered through the question of perception. As Diana Coole outlines, a phenomenological approach shows that the body has an active role in the generation of meaning through perception, which renders perception rather than reason as primary (Coole 129). Drawing upon the ideas of Maurice Merleau-Ponty, Coole remarks that the perceiving body is integrated into a configuration that is already “patterned”. Thus, “each perception is an interpretation that ‘re-enacts on its own account the birth of intelligence and has some element of creative genius about it’” (Merleau-Ponty 37; Coole 129). James Gibson’s concept of ‘the field of safe travel’ describes the role perception plays in the construction of the successful driver. It consists, at any given moment, “of the field of possible paths which the car may take unimpeded” (Gibson 120; Dant 63). Waymo’s
22
perceptual engagement with its environment occurs via its laser sensing technology called LIDAR (Figure 2). LIDAR emits pulses of infrared light to detect and analyze the vehicle’s position in relation to the surrounding objects. Within this visual field Waymo sees, predicts and plans a safe path for the vehicle; the different colored box models and trajectories illustrate the relationship between perception, intention and action. Furthermore, LIDAR does not “just [identify and predict objects’ behavior] for the objects you and I can see. It can do that for things up to three football fields away” (Waymo.com/Tech). As I see it, this illustrates Tim Dant’s argument about the driver-car assemblage, where “neither the human driver nor the car acting apart could bring about the types of action that the assemblage can” (Dant 62). Furthermore, it implies an idea of extended or heightened perception by technological proxy. It extends the driver-car’s visual field to anticipate potential obstructions –approaching cyclist, pedestrian crossing the road– within a radius that is beyond the human driver’s perceptual skills. As Luciano Floridi puts it:
The point is that our smart technologies – also thanks to the enormous amount of available data and some very sophisticated programming – are increasingly able to deal with more tasks
better than we do, including predicting our behaviours. So we are not the only agents able to perform tasks successfully. (Floridi 2017)
As noted by scholars such as Nagy and Neff, for example, ideas of proxy agency can be extended to account for technological agency to rethink how human and technological agencies are co-constituted (Nagy & Neff 4917). As such, the notion of proxy agency can be used to consider how technological systems function as moral agents, through algorithmic modeling of human morals. However, proxy agency doesn’t describe the networked condition of agency beyond singular agentic relationships. In its approach to the agentic relationship it creates a clear distinction between the object and the subject, which is treated as the primary agent of any ensuing action. Therefore it cannot fully describe the enmeshment of human and nonhuman agencies. As Edwin Sayes argues, we can only speak of the agency of a particular nonhuman if we were to “ignore all the humans and other nonhumans that are lined up behind it” in order to ascribe that nonhuman its continued agency (Sayes 143). In addition, the notion of proxy agency leans too heavily in favor of casting the human as the primary source of agency. The driver-car assemblage of Waymo becomes imbued with human desire and intentionality through the design, functionality and use of Waymo. However, the human agencies inherent to Waymo work in symbiosis with nonhuman agencies, creating a rhizomatic agentic relationship, where both the material and the discursive are at play. Thus, the concept of proxy agency leaves the question of a more affirmative understanding of moral agency open.
MORAL AGENCY IN NETWORKS
In favor of more symmetrical conceptualization of agency is Bruno Latour, whose Actor-Network Theory (ANT) is associated with many theories on the human-nonhuman relationship and the redistribution of agency in networks. What is significant about ANT is its
acknowledgement of material objects as actors in networks, including, for example, aircraft (Law and Callon 1992), transit systems (Latour 1996) or electric cars (Callon 1986). ANT conceptualizes agency as a symmetrical power to influence human or nonhuman subjects or objects. As such, it works towards the idea of a collective of humans and nonhumans, where agency is no longer seen as only comprising human subjectivity and autonomy (Radomska 94). However, as Tim Dant outlines, it does not tell much about the “lived nature of human beings and objects” (Dant 70). Nevertheless, ANT serves to foreground a key critical issue inherent to my argument here, which is that of networks.
As illustrated by Martin Müller and Carolin Schurr, the ideas and conceptual vocabulary intrinsic to ANT can be enriched by combining them with assemblage thinking (Müller & Schurr 217). The concept of the assemblage as originally envisioned by Deleuze and Guattari encompasses the relationships and configurations between materials, locations, spaces, discourses and bodies that “co-constitute the choreographies of action” (Shildrick 18). As I see it, assemblage thinking corresponds with Barad’s intra-action model since it doesn’t presume the separateness of objects or things, but thinks about the social world through ideas of fluidity and configuration. Furthermore, assemblage thinking shows openness to the idea of change through unpredictable rupture, towards the event (Müller & Schurr 226). As Alecia Jackson and Lisa Mazzei outline, “an assemblage isn’t a thing–it’s a process of making and unmaking the thing” (Jackson & Mazzei 1).
The conceptual openness of the assemblage supports the idea of Waymo as a disruption within the driver-car. As Tim Dant outlines, the driver-car is about a set of social actions enabled by the collaboration between humans and machines. The networked condition of Waymo extends the array of social actions that the driver-car assemblage can perform by interlinking bodies, systems of knowledge and technology into vast networks. Take, for instance, Waymo’s Carcraft software, which functions as the simulation technology for Waymo’s autonomous vehicles. The world of Carcraft consists of a highly-detailed representation of the world of virtual cars, lanes, curbs, traffic lights, intersections and roundabouts. Within Carcraft, 25 000 Waymo vehicles drive 8 million miles every day, testing different driving scenarios based on replicas of every real-world mile the Waymo vehicles have driven on public roads and on the driving test track (Figure).13 Furthermore, Carcraft creates an exhaustive amount of different versions of the same scenarios, adjusting everything from the timing of traffic lights to the speed of oncoming vehicles, cyclists or
pedestrians to the weather conditions of the drive. The dense amount of data that Carcraft 13 For more detail, see : https://medium.com/waymo/simulation-how-one-flashing-yellow-light-turns-into-thousands-of-hours-of-experience-a7a1cb475565
26
Figure 3: Carcraft’s real and virtual view of an intersection in Phoenix, Arizona. Image: Waymo.
processes on a constant basis is distributed into the networked virtual-material fleet of Waymo vehicles.
As far as I am concerned, Carcraft signifies the idea of the infinitely malleable assemblage, a scatter of “meanings, representations, practices and subjectivations, offering new ‘becomings’” (Youdell & Armstrong, 145). The multiplicity of the virtual-material practices creates, to put it in Deleuzoguattarian terms, a “centre of vibrations”, which stands for “the configuration, the constellation of an event to come” (Deleuze & Guattari 32). As such, Carcraft alludes towards the idea of the co-constitution of the material-virtual arrangements within a constant reconfiguration of the world. It corresponds with Karen Barad’s idea of material phenomena, which is defined by “the ontological inseparability/entanglement of intra-acting agencies” (Barad Meeting 139). Furthermore, Carcraft illustrates the productive relations of the assemblage that material-virtual bodies enter into through intra-action. Carcraft and the real world shape and reshape one other through a network of virtual simulations and material technology. As outlined in an article about Waymo in the Atlantic, this process is described as follows:
A memory of a roundabout in Austin becomes a piece of [Castle becomes a self-driving car data log becomes a Carcraft scenario becomes a web of simulations becomes new software that finally heads back out on a physical self-driving car to that roundabout in Texas. […] Carcraft should be remembered as a virtual world that had an outsized role in reshaping the actual world on which it is based. (The Atlantic)
The nonhuman agencies inherent in Carcraft illustrate Barad’s claim that nonhumans “partake in the world’s active engagement in practices of knowing”. In this sense, intelligibility is seen as “an ontological performance in its ongoing articulation”, a process of differential becoming; an assemblage. As such, Carcraft demonstrates the Baradian idea of knowing as a matter of being and intra-acting. For Barad, the production of meanings is
always entangled with the production of material bodies and boundaries. (Barad, Meeting 149). With this in mind, I would like to argue that a crucial concept in terms of Waymo’s engagement in practices of knowing is that of experience. The millions of virtual and real-life miles that Waymo drives every day on public roads and in simulation are done within the intent of “building the world’s most experienced driver” (Waymo.com). Waymo’s technology relies on the exhaustive amount of simulations that consist of digital recreations of thousands of variations of any single scenario. As such, as I see it, Waymo’s agential functioning can be accounted as “differential responsiveness and accountability as part of a network of performances” (Barad, Meeting 149). Resonant with this argument is Luciano Floridi’s theorization of the relationship between intelligibility and memory. According to Floridi, algorithms and datasets can outperform intelligence, which renders the questions of technology’s mental or emotional states, consciousness or intentions secondary (Floridi 2016). In fact, Waymo posits experience at the heart of its technology: “What makes everything you can see right now possible is experience” (Waymo.com/Tech). As I see it, the questions of memory and experience are thus crucial in terms of developing our understanding of the moral agency of technological systems.
In conclusion, this chapter has outlined the disruption Waymo produces within the traditional driver-car assemblage. The nonhuman agencies inherent to Waymo produce a new kind of moral subject that subverts the human/nonhuman, mind/body and virtual/actual dichotomies. Self-driving technology does not transform computer systems into explicit moral agents, but connects technological components into vast virtual-material networks, where they function as part of morality. And as they do so, their ethically significant behavior is conditioned by the technological affordances that result from human intentionality and desire. The values and rules programmed into Waymo affect how the vehicle sees, identifies, predicts and plans its actions within the material environment. Moreover, the moral system
incorporated into Waymo’s technology is realized in conjunction with other human and nonhuman bodies. Thus the boundaries and meanings of Waymo become constituted in an entanglement with other intra-acting agencies. As such, the moral subject becomes an endlessly malleable assemblage, encompassing both human and nonhuman agencies.
CHAPTER 2: IMAGINARY AFFORDANCES AND AFFECTIVE AESTHETICS
As stated in the previous chapter, in the currently shifting technological landscape, a representational logic for moral agency can no longer account for a comprehensive understanding of the role A. I. systems play within society. A posthumanist performative approach, such as that of Barad, permits an entry point in order to examine how Waymo functions as a part of moral agency as a result of agential entanglement comprising material bodies, software and space. Waymo’s moral agency emerges within the driver-car assemblage in the form of particular affordances relating to the distribution of cognition of its A. I. system and other technological and technical features that determine Waymo’s agentic properties and boundaries.
In this chapter I will turn to focus on how different expectations and capacities relating to Waymo’s moral agency are further established affectively and discursively. Waymo’s online presence serves to reflect various critical issues I would like to address in order to develop an adequate analysis on how the affective aesthetics of Waymo relates to its moral agency. For example, visitors to the Waymo website are introduced to information on the vehicles’ technology, the journey of Waymo as a company and information about the ongoing driving tests, as well as additional information about how to join the Early Riders program, which is a public trial of Waymo’s autonomous vehicles. The website contains text, images and video material that reflect Waymo’s overall marketing brand and suggest how the company relates to the web and the average visitor. The issue of the moral agency of autonomous vehicles– that has, as previously stated, been a prominent feature in the public discussion surrounding self-driving technology–is, however, not outwardly addressed. The issues concerning the moral status of Waymo’s self-driving technology appear somewhat abstractly, being concealed within the textual and visual descriptions of the technology. Take this following passage, for example:
We’re building a safer driver that is always alert and never distracted. Our fully self-driving technology will handle all the driving so you can go from door to door without taking the wheel. This will deliver the biggest impact on improving road safety and mobility for everyone (Waymo.com/tech).
This introduction foregrounds Waymo’s fully autonomous vehicle technology with an emphasis on the safety and efficiency of the vehicle as a vantage point of Waymo’s technology. What underlies this description is, yet again, the premise of the unreliability of the human driver and the rationale behind the decision to remove all human input from the driving task. Indeed, what follows this description as you scroll down the page is data concerning ‘fatalities associated with human choice or error’ that, according the company’s statistics, are featured in 94% of vehicle crashes in the United States (Waymo.com/tech). The introduction signifies an integral part of Waymo’s mission to reduce the amount of traffic fatalities each year14; Waymo’s autonomous vehicles are designed to save lives. Furthermore, what is significant here is that most of the time Waymo is personified as ‘the driver’; ‘the safer driver’, ‘the most experienced driver’ and so forth. In addition, these personifications emphasize accessibility and equality: Waymo is the self-driving technology for “everyone”. These discursive mechanisms that power Waymo further reflect a new kind of agentic subjectivity within the driver-car assemblage discussed in the previous chapter.
Having discussed Waymo as a technological system that is afforded moral agency through its technological capacities that emerge within the human-machine intra-actions, I wish to focus on Waymo’s imagined affordances. The strength of Gibson’s concept of affordance introduced in the previous chapter lies in the way that it “establishes the properties of material things in relation to a particular species” (Dant 65). While providing a productive entry point for examining the physical capacities of material objects in the ‘real’ world, the notion of affordance does not fully explain how meanings and properties of objects are 14 For a more detailed account, see: http://x.company/waymo/
generated socially and discursively. In order to gain further understanding on how the boundaries and properties of Waymo’s moral agency are being constituted, in this chapter I wish to specifically focus on Waymo as a discursive phenomenon (although always intertwined with the material).
Thus this chapter aims to develop a Post-Gibsonian analysis of Waymo’s affordances that, as I see it, result from a tandem process of affect, mediation, perception and cognition. To begin, I will provide an analysis of how Waymo’s affordances emerge from human expectations, perceptions and attitudes by applying a reconstructed interpretation of the concept of affordance, provided by Communication Studies theorists Peter Nagy and Gina Neff (Nagy & Neff 1). Through engaging with the concept of ‘imagined affordance’, this chapter discusses how mediation, materiality and affect work to shape the discourse surrounding Waymo. Keeping up with Barad’s performative approach, I will focus on how the discursivity of Waymo emerges from human/nonhuman intra-actions. Moreover, I will outline the discursive methods through which Waymo’s moral status and agency are inconspicuously, yet strategically, prompted. Finally, this chapter will argue that the imagined affordances of Waymo must be taken into account in the consideration of how its moral agency materializes through the human/nonhuman relationship. Thus, this chapter will extend my previous analysis of Waymo as a component of moral agency.
FROM IMAGINED AFFORDANCES TO MATERIAL CONFIGURATIONS
As I see it, Waymo’s moral agency emerges not only from what it can do, but what people imagine it can do. Thus, in short, imagination plays a crucial part in how Waymo is designed, perceived and used. Rethinking the Gibsonian ‘affordance’, Peter Nagy and Gina Neff aim to trouble the use of affordance as static and human-dependent concept. Even though the concept of affordance goes a long way towards describing the complex
nonhuman relationship by resisting both the mind/body dualism and a separate cognitive phase where a certain action is programmed, it has still been seen as relying too heavily on the idea of human as the main source of agency (Dant 66). Nagy and Neff extend this criticism by claiming that especially within Communication theory, the concept of affordance is often synonymous with “technological features”, which “doesn’t incorporate the material, mediated and emotional aspect of human-technology interaction” (Nagy & Neff Imagined 2). As a corrective, Nagy and Neff propose the concept of “imagined affordances”. That is, imagined affordances that emerge from users’ and designers’ imaginations: their perceptions, attitudes and expectations for technology “that are not fully realized in conscious, rational knowledge” (Nagy & Neff Imagined 1). As such, imagined affordances reflect the potential relationships between humans and technology; they are “properties of the world” and come into being through the relation between ‘the material’ and ‘the perceptual’ (Nagy & Neff Imagined 3).
Following Nagy and Neff, I am inclined to see users’ and designers’ perceptions, fears, attitudes and expectations play a significant role in the ways the meanings of technology are configured and reconfigured within human-technology relationships. As I see it, desire, imagination, perception and affect bridge the human use of technology with the materiality and functionality of technologies. For instance, Waymo’s journey from semi-automated cars to fully autonomous vehicles reflects not only technological development, but also conscious decisions in their design. For instance, Waymo’s experiences with semi-autonomous vehicles showed that as soon as humans believe that technology has the slightest amount of autonomy, they overestimate its capabilities, which made the company move onto fully autonomous design (Madrigal 2017). This demonstrates a tension between perception, design and use, powered by human imagination. Imagination, as argued by Gibson, can be considered as an extension of perceptual knowledge (Gibson 69). Furthermore, imaginative
processes are considered to affect how people perceive technological affordances (Boschker, Bakker, & Michaels 782). Therefore, it can be argued that imagination and agency exist within an entanglement. As Koestler suggests, “ […] all our perceptions are coloured by imagination. Thus the difference between “real” and “imaginary” environments become a matter of degrees” (Koestler 103). Furthermore, this illustrates a looped relationship between desire, imagination and technology. As Robert Pepperell and Michael Punt argue, human imagination begets desire that begets technology, which in turn propagates further desire (Pepperell & Punt 35). In this light, Waymo can be considered by being informed by imagined affordances for its technology: perceptions and intentions of designers, the expectations and perceptions of the users, and the materiality of the vehicle. For instance, this tension is illustrated in a video clip found on the Waymo website. The video shows people riding around in a Waymo vehicle, observing their surroundings and people’s reactions to the vehicle. In one of the conversations between the passengers, a man says: “[…] She was like, “is there no one driving that vehicle?””, to which his fellow passenger replies, laughing: “I knew it! I was expecting it!”. Thus, the video describes the expectations and perceptions of the users. Seeing a regular looking vehicle with no one behind the steering wheel disrupts the common imagery of the driver-car assemblage that connotes perception. Furthermore, it illustrates the designers’ expectations of the “imagined users”. This describes a process of “encoding imagination”, which assumes an “an act of extended consciousness”, a “corollary process of decoding”, (Pepperell & Punt 68), where the imaginations of the designers and the users are conjoined across time and space. In Waymo’s case, this describes the entanglement of imagined and real affordances.
As Nagy and Neff argue, imagined affordances can include expectations and beliefs of users, “whether or not they are “true” or “right”” (Nagy & Neff Imagined 4). They do not need to be realized in conscious knowledge to be “concretized or materialized in
technical systems” (Nagy & Neff Imagined 2). Furthermore, the materialization of imagined affordances alludes to the relation of matter and meaning, and the possibility of thinking about knowing as a matter of intra-acting, of “being in the world” (Barad, Meeting 149). Moreover, this entanglement describes how material-discursive practices emerge from intra-actions between material bodies. The meanings, boundaries and affordances of Waymo emerge from the complex intra-actions between Waymo’s designers, programmers and users, as well as the material vehicles themselves, being both “environmental and perceptual, both conceptual and imagined” (Nagy & Neff, Imagined 3). Furthermore, these multiple material-discursive practices determine Waymo’s position as a component of moral agency. As Nagy and Neff point out, the power of imagination in reshaping technologies cannot be understated (Nagy & Neff Talking 4926).
What I find significant here, specifically, is how material-discursive technologies shape human imagination and perception; in essence, our way of thinking. Rapidly shifting and changing, and constantly unfolding, new technologies remind me of the Deleuzoguattarian idea of “centers of vibrations”, concepts “virtually containing the possibility of an event to come” (Radomska 106). As a fully autonomous vehicle, Waymo subverts the singularity and exceptionality of moral agency as an exclusively human concept and reveals the potential of “becoming” through “differing”. If we look at moral agency as a philosophical concept that is “the configuration, the constellation of an event to come”, it becomes a concept already containing the possibility to think it through and move further; “to push [it] to the limit and go beyond [it]”. (Radomska 99). With this in mind, I will turn to examine the aesthetic and affective dimensions of Waymo, and their entanglement with discourse.
POSTHUMAN AESTHETICS AND DESIGN
The Waymo website reflects the overall aesthetic of Waymo as a company. The website’s design communicates the usability, narrative and visuality of Waymo. The visual of the website is crisp and bright, and carries a resemblance to Google products. The functionality of is clear, simple and easy to navigate. The accessibility of the website is coded and constructed in relation to an average visitor, which is reflected, for example, on the level of narrative hierarchy. The different sections of the website follow an alternating order of text and visual aid, such as illustrations of traffic, video clips, and pictures of the vehicle. This creates a layered, hierarchical ‘game-like’ narrative that I would describe as playful, fun, and welcoming. Furthermore, the functionality of the website constructs a narrative loop, where the user is directed onto the next section of the website instead of having to ‘seek’ information themselves. As such, it also creates a feeling of informational meticulousness, even though, in terms of informative content, the Waymo website is relatively scarce. Furthermore, the quality of the provided content appears somewhat undetailed. For instance, take the passage of how Waymo describes how its technology works: “Our vehicles have sensors and software that are designed to detect pedestrians, cyclists, vehicles, road work and more from up to three football fields away in all 360 degrees.” (Waymo.com/tech). Thus, it appears that Waymo relies on its design’s capability of building an understanding of its technology.
There is an increasing amount of evidence to show that aesthetics is one of key factors that shape the engagements of humans with technologies (Umemuro 373). Furthermore, the boundaries between art and technology are becoming increasingly blurred through the aesthetic design of technological commodities. This makes aesthetic an important category to explore in relation to Waymo. Following Deleuze’s line of thought, Simon O’Sullivan argues that the aesthetic can be considered as an “affirmative actualization of the virtual”, consisting of a realm of affects that materialize through a creative engagement. Furthermore, as
O’Sullivan argues, this renders the aesthetic its ethical imperative (O’Sullivan 129). As Guattari argues, the ethico-aesthetic paradigm is inherently a matter of the production of subjectivity. For Guattari, individuals create “new modalities of subjectivity” through a pragmatic and aesthetic configuration. (Guattari Chaosmosis 7). In this light, aesthetics become an active creator of new concepts and moral configurations.
With this in mind, I would like to consider Waymo’s design on the material level of two of the Waymo’s vehicle designs: The Firefly and the Chysler Pacifica Hybrid minivan. The Firefly (Figure 5) was a round-topped two-seat Waymo prototype that was introduced to the public in 2014. The design of The Firefly pushed Waymo into new territory, which is referred to on the website; “We explored what fully self-driving cars could be like by designing a new reference vehicle. […] These cars had custom sensors, computers, steering and braking, but no steering wheel or pedals.” (Waymo.com/journey). The design of the vehicle thus differed significantly from the traditional motor car. The Firefly had simple round contours and minimalistic ornamentation, which made it appear soft and sort of “doughy”. Furthermore, looking at the vehicle from the front of the car, its front lights and
