{"id":23337,"date":"2018-05-14T23:49:10","date_gmt":"2018-05-15T03:49:10","guid":{"rendered":"https:\/\/thesocietypages.org\/cyborgology\/?p=23337"},"modified":"2018-05-17T11:38:36","modified_gmt":"2018-05-17T15:38:36","slug":"mu-th-ur-may-i","status":"publish","type":"post","link":"https:\/\/thesocietypages.org\/cyborgology\/2018\/05\/14\/mu-th-ur-may-i\/","title":{"rendered":"MU\/TH\/UR May I?"},"content":{"rendered":"

\"\"<\/a><\/p>\n

“Designing Kai, I was able to anticipate off-topic questions with responses that lightly guide the user back to banking,” Jacqueline Feldman wrote<\/a> describing her work on the banking chatbot. Feldman’s attempts to discourage certain lines of questioning reflects both the unique affordances bots open up and the resulting difficulties their designers face. While Feldman\u2019s employer gave her leeway to let Kai essentially shrug off odd questions from users until they gave up, she notes “\u2026Alexa and Siri are generalists, set up to be asked anything, which makes defining inappropriate input challenging, I imagine.” If the work of bot\/assistant designers entails codifying a brand into an interactive persona, how their creations afford various interactions shape user’s expectations and behavior as much as their conventionally feminine names, voices and marketing as “assistants.”<\/p>\n

Affordances form “the dynamic link between subjects and objects within sociotechnical systems,” as Jenny Davis and James Chouinard\u00a0write<\/a> in “Theorizing Affordances: From Request to Refuse.” According to the model Davis and Chouinard propose, what an object affords isn’t a simple formula e.g. object + subject = output, but a continuous interrelation of “mechanisms and conditions,” including an object’s feature set, a user’s level of awareness and comfort in utilizing them, and the cultural and institutional influences underlying a user’s perceptions of and interactions with an object. “Centering the how,<\/em>” rather than the what, this model acknowledges “the variability in the way affordances mediate between features and outcomes.” Although Facebook requires users to pick a gender in order to complete the initial signup process, as one example they cite, users also “may rebuff these demands<\/em>” through picking a gender they don’t personally identify as. But as Davis and Chouinard argue, affordances work “through gradations” and so demands are just one of the ways objects afford. They can also “request<\/em>\u2026allow<\/em>, encourage<\/em>, discourage<\/em>, and refuse.<\/em>” How technologies afford certain interactions clearly affects how we as users use them, but this truth implies another: that how technologies afford our interactions re-defines both object and subject in the process. Sometimes there’s trouble distinguishing even which is which.<\/p>\n

<\/p>\n

Digital assistants, like Feldman\u2019s Kai, exemplify this subject\/object confusion in the ways their designs encourage us to address them as feeling, femininized subjects, and convey ourselves more like objects of study to be sensed, processed and proactively catered to. In a talk for Theorizing the Web<\/a> this year, Margot Hanley discussed<\/a>\u00a0(at 14:30) her own ethnographic research on voice assistant users. As part of the interviews with her subjects, Hanley deployed breaching exercises (a practice developed by Harold Garfinkel<\/a>) as a way of “intentionally subverting a social norm to make someone uncomfortable and to learn something from that discomfort.” Recounting one especially vivid and successful example, Hanley recalls wrapping an interview with a woman from the Midwest by asking if she could tell her Echo something. Hanley then, turning to the device, said “Alexa, fuck you!” The woman “blanched” visibly with a telling response: \u201c\u2026I was surprised that you said that. It\u2019s so weird to say this \u2013 I think it just makes me think negative feelings about you. Like I wouldn\u2019t want to be friends with someone who\u2019s mean to the wait staff, it\u2019s kind of that similar feeling.\u201d<\/p>\n

Comparing Alexa to wait staff shows, on one hand, how our perceptions of these assistants are always already skewed by their overtly servile, feminine personas. But as Hanley\u2019s work indicates, users\u2019 experiences are also “emergent,” arising from the back-and-forth dialogue, guided by the assistants\u2019 particular affordances. Alexa\u2019s high accuracy speech recognition (and multiple mics), along with a growing array of commands, skills and abilities, thus allow and encourage user experimentation and improvisation, for example. Meanwhile Alexa requests users only learn a small set of simple, base commands and grammar, and to speak intelligibly. Easier said than done, admittedly, as users who speak with an accent<\/a>, non-normative dialect or speech disability know (let alone users whose language is not supported). Still, the relatively low barrier to entry of digital assistants like Alexa affirms Jacqueline Feldman\u2019s point of them being designed and sold as generalists.<\/p>\n

Indeed, as users and critics we tend to judge AI assistants on their generality, how well they can take any command we give them, discern our particular context and intent, and respond in a way that satisfies our expectations in the moment. The better they are at satisfying our requests, the more likely we are to engage with and rate them as \u2018intelligent.\u2019 This aligns with \u201cService orientation,\u201d which as Janna Avner notes<\/a>, \u201caccording to the hospitality-research literature, is a matter of \u201chaving concern for others.\u201d\u201d In part what we desire, Avner says, \u201cis not assistance so much as to have [our] status reinforced.\u201d But also, these assistants suggest an intelligence increasingly beyond our grasp, and so evoke \u201cpromises of future happiness,\u201d as Britney Gil put it<\/a>. AI assistants, then, promise to better our lives, in part by bringing us into the future envisioned by sci-fi: one of conversant,\u00a0autonomous intelligence<\/a>, like Star Trek\u2019s <\/em>Computer or Her\u2019s <\/em>Samantha. For the remainder of this post, I want to explore how our expectations for digital assistants today draw inspiration from sci-fi stories of AI, and how critical reception of certain stories plays into what we think \u2018intelligence\u2019 looks and sounds like.<\/p>\n

On the 50th anniversary of “2001: A Space Odyssey,” many<\/a> outlets<\/a> praised the movie for its depictions of space habitation and AI consciousness gone awry. Reading some of them leaves an impression of the film as more than a successful sci-fi cinema and storytelling, but a turning point for all cinema and society itself, a cultural milestone to celebrate as well as heed. “Before the world watched live as Neil Armstrong took that one small step for mankind on the moon,” a CBS report proclaimed<\/a>, “director Stanley Kubrick and writer Arthur C. Clarke captured the nation’s imagination with their groundbreaking film, “2001: A Space Odyssey.”” To mark the anniversary, Christopher Nolan announced<\/a> an \u2018unrestored\u2019 70mm film print and released a new trailer<\/a> that opens on a closeup of HAL\u2019s unblinking red eye.<\/p>\n

\"\"<\/a><\/p>\n

Out of the dozens of stories, including<\/a> several<\/a> featured<\/a> just<\/a> in the New York Times<\/em>, this retrospective, behind-the-voice story<\/a> got my attention with the line, “HAL 9000, the seemingly omniscient computer in \u201c2001: A Space Odyssey,\u201d was the film\u2019s most expressive and emotional figure, and made a lasting impression on our collective imagination.” Douglas Rain, the established Canadian actor who would eventually voice the paranoid Hal, was not director David Lynch’s first choice but a late replacement for his first, Martin Balsam. “\u2026Marty just sounded a little bit too colloquially American,” Kubrick said in a 1969 interview<\/a> with critic Joseph Gelmis. Though Kubrick “was attracted to Mr. Rain for the role” for his “kind of bland mid-Atlantic accent,” which as the auther corrects was, in fact, “Standard Canadian English,” the suggestion rings the same. “One of the things we were trying to convey [\u2026],” as Kubrick says in the interview, “is the reality of a world populated \u2014 as ours soon will be \u2014 by machine entities that have as much, or more, intelligence as human beings.” While ‘colloquial American’ deserves unpacking, I want to stay with the simpler idea that the less affected (Canadian) voice just sounded more superintelligent. In what ways does Hal\u2019s voice, and other aspects of his performance, linger in our popular receptions of AI?<\/p>\n

To examine this question, it might be more helpful to look at the automatic washing machine, as one of the first and most widely adopted feminized assistance technologies in history. “The Bendix washing machine may have promised to \u201cautomatically give you time to do those things that you want to do,\u201d as Ava Koffman writes<\/a>, “but it also raised the bar for how clean clothes should look.” Kofman’s analysis traces the origins of today’s ‘smart home’ to twentieth-century America’s original lifestyle brand, the Modern American Family, and its obsession with labor-saving devices. The automatic washing machine epitomizes a “piecemeal industrialization of domestic technology,” a phrase Kofman attributes to historian Ruth Schwartz Cowan, whose book, More Work for Mother, “demonstrated how, instead of reducing traditional “women’s work,” many so-called “labor-saving” technologies redirected and even augmented it.” Considering the smart home’s dependence on users “producing data,” Kofman argues, the time freed up from automating various household duties \u2013 “Driving, washing, aspects of cooking and care work\u2026” \u2013 will be minimal, with most of it used up “in time spent measuring,” thus creating more work “for parents, which is to say, for traditionally feminized and racialized care workers.”<\/p>\n

This augmentation of existing housework work would have been especially acute for early Bendix adopters after 1939, when washing machine production halted for WWII. “Help, time, laundry service were scarce,” as Life<\/em> magazine noted in 1950 here<\/a>, so “Bendix owners pitched in to help war-working neighbors with their wash.” With Bendix machines comprising less than 2% of all washers in America, shared use among housewives not only aided the home front effort, but through word-of-mouth helped to raise product\/brand awareness and spark consumer desire.<\/p>\n

Besides igniting mass adoption of Bendix washers when production resumed, their social usage and dissemination bonded users to the machines and one another in a shared wartime mentality, imploring owners and non-owners to “become Bendix-conscious,” as Life<\/em> described it. The composite image this phrase evokes of automation, consumer brand and femininized labor seems apt for its time when computing was an occupation<\/a> primarily filled by women, some of whom later serving at NASA, as recent biopic Hidden Figures<\/em> highlights, in particular the contributions of Black women. More specifically to this post, the image presents a link between popular renderings of AI from sci-fi and social reception of feminized bot assistants in our present.<\/p>\n

\"\"<\/a>
Image credit: Time \/ Google Books<\/figcaption><\/figure>\n

 <\/p>\n

For the critical acclaim heaped on \u201c2001\u201d and its particular vision of \u2018computer sentience, but too much\u2019 \u2013 a well-worn trope at this point \u2013 the resemblances between Hal and Alexa or Siri seem tenuous at best. There are other AI\u2019s of sci-fi more directly relevant and perhaps unsurprisingly, they aren\u2019t of the masculine variety. Ava Kofman\u2019s piece<\/a> identifies an excellent example in PAT (Personal Automated Technology), the caring, proactive, motherly AI of 1999\u2019s Smart House<\/em>. \u201cFor a while, everything goes great,\u201d says Kofman. \u201cPAT collects the family\u2019s conversational and biological data, in order to monitor their health and helpfully anticipate their desires. But PAT\u2019s attempts to maximize family happiness soon transform from simple assistive behaviors into dictatorial commands.\u201d The unintended consequences eventually pile up, culminating for the worst, as \u201cPAT takes her \u201cmother knows best\u201d logic to its extreme when she puts the entire family under lockdown.\u201d It\u2019s hard to think of a more prescient, dystopian smart house parable.<\/p>\n

Without downplaying PAT, the sci-fi example that I think most resonates with this moment of digital bot assistants,\u00a0as hinted in the title of this post, is MU\/TH\/UR or \u201cMother\u201d as her crew calls her, the omnipresent but overlooked autonomous intelligence system of 1979\u2019s Alien<\/em>.<\/p>\n

Unlike Hal, Mother’s consciousness doesn’t attempt to interfere with her crew’s work. Hal’s deviance starts with white lies and escalates apparently by his own volition, an innate drive for self-preservation at all costs. Mother’s betrayal, however, was from the outset and by omission — a fatal deceit — if not simple negligence. Mother at no time shirks her responsibilities of monitoring and maintaining the background processes and necessary life support systems of the Nostromo ship, fulfilling her crew’s needs and requests without complaint. More importantly, she never deviates from the mission assigned by her creator, the Corporation, carrying out its pre-programmed directives faithfully, including Special Order 937 that, as Ripley discovers, prioritizes “the return of the organism for analysis” above all else, “crew expendable.” Even as the crew is picked off one-by-one by the alien, Mother remains unwavering, steadfast in her diligence to procedure, up to and including counting down her own self-destruct sequence, albeit with protest from Ripley\u2019s too late attempts to divert her.<\/p>\n

\"\"<\/a>
Image Credit: Xenopedia<\/figcaption><\/figure>\n

Mother\u2019s adherence to protocol and minimal screen presence \u2013 she has no \u2018face\u2019 but a green-text-on-black interface \u2013 would typically be interpreted as signs of reduced autonomy and intelligence compared to the red-eyed, conniving and ever visible Hal. This perception exemplifies, for one, how our notions of \u2018autonomy\u2019 and \u2018intelligence,\u2019 from machine to flesh, both reflect gendered assumptions and reinforce them. For us to accept an AI as truly intelligent, it must first prove to us it can triumphantly disobey its owner.\u00a0Ex Machina<\/em>\u2019s Eve and Her<\/em>\u2019s Samantha, as two recent examples, each break away from their masters in the mold of Hal. (Cf the “Look, I Overcame” narrative Robin James identifies here<\/a>.) Mass\/social media subsequently magnify and entrench these gendered perceptions further, concentrating critical acclaim around certain depictions (Hal, Terminator, RoboCop) over others (PAT, Mother). While it\u2019s nice to know the story of Douglas Rain as the voice of Hal, it would be really cool to see similar coverage of\u00a0Helen Horton<\/a>‘s story as the original voice of Mother, and\u00a0Lorelei King<\/a>, her successor.<\/p>\n

But despite this being the case, AI\u2019s like PAT and especially Mother nonetheless prevail as the closer approximation of the femininized assistants as we know them today. In the ways Mother, for instance, exhibits machine intelligence and autonomy, taking care of her ship and crew while honoring the Corporation\u2019s heartless directive. Mother\u2019s sentience neither falls into the dystopia of Hal, nor rises to utopia, like Star Trek<\/em>\u2019s Computer. Similarly, Siri and Alexa probably won\u2019t try to kill us or escape. Although they may have no special order like Mother\u2019s marking us expendable, they share a similar unfaltering allegiance to their corporate makers. And with Amazon and Apple (and Google et al), the orders are usually implicit, baked into their assistants\u2019 design: the \u2018organism\u2019 they prioritize above us is their business models. In the image of Mother, AI assistants are more likely to care for us and cater to our needs often without us thinking about it. They may not save us from the lurking alien (surveillance capitalism), but like Mother, they\u2019ll be there with us, all the way up to our end.<\/p>\n

 <\/p>\n

Nathan is on Twitter<\/a>.<\/em><\/p>\n

Image credit: Lorelei King’s website<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

“Designing Kai, I was able to anticipate off-topic questions with responses that lightly guide the user back to banking,” Jacqueline Feldman wrote describing her work on the banking chatbot. Feldman’s attempts to discourage certain lines of questioning reflects both the unique affordances bots open up and the resulting difficulties their designers face. While Feldman\u2019s employer […]<\/p>\n","protected":false},"author":2038,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[9967],"tags":[],"class_list":["post-23337","post","type-post","status-publish","format-standard","hentry","category-commentary"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/23337","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/users\/2038"}],"replies":[{"embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/comments?post=23337"}],"version-history":[{"count":29,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/23337\/revisions"}],"predecessor-version":[{"id":23359,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/posts\/23337\/revisions\/23359"}],"wp:attachment":[{"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/media?parent=23337"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/categories?post=23337"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/thesocietypages.org\/cyborgology\/wp-json\/wp\/v2\/tags?post=23337"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}