Download as pdf
Download as pdf
You are on page 1of 16
AM> Artie ‘ASSOCIATION no a SAGE Bad News? Send an Al. Good News? Send a Human ‘Aaron M. Garvey, TaeWoo Kim, and Adam Duhachek Abstract The present research demonstrates how consumer responses co negative and postive offers are influenced by whether the administering marketing agent is an artical iceligence (Al) or a human. In the case of a product or service offer that Is Worse than expected, consumers respond better when dealing with an Al agent in the form of Inreased purchase likelihood and satisfaction. In contrast, for an offer vhat is better than expected, consumers respond more poslvely to a human agent. “The authors demonstrare that Al agents. compared with human agents, are perceived eo have weaker intentions when admin- iscering offers, which accounts for this elect. Thats, consumers infer that Al agent lack selfish intentions in the ease ofan offer ‘that favors the agent ancl lack benevolent intentions in the case of an offer that favors the customer, thereby dampening the ‘ecremity of consumer responses. Moreover, the auchors demonstrate a moderating effect. such chat marketers may anthropo- rmorphize Al agents to strengthen perceived intentions, providing an avenue to receive due cred from consumers when the agent provides 2 betcer offer and mitigate blame when ic provides a worse offer. Povenal ethieal conceens with the use of Al 10 bypass consumer resistance to negative offers are discussed Keywords aril inelgence, anchropomorphism, algorithm, robots, technology, pricing, intentions, expectations, satisfaction (Online supplement: hexpsildoiorg/10.1177/0022242921 1066972 Marketing managers currently find themselves in a period of technological transition, wherein artificial intelligence (AN) agents are increasingly viable replacement options for human representatives in administering. product and service offers directly to customers. Al agents have been adopted across a broad range of sonsumer domains to handle both face-to-fase and remote customer transactions (Davenport et al, 2020; Haris, Kimson, and Schwedel 2018; Huang and Rust 2013; Wine et al. 2018), ranging ftom traditional retail and travel (Mondo etl, 2019) to ride and residence sharing (Hughes etal 2019) and even legal and medical services (Esteva etal 2017; Turner 2016). Given Al agents’ advanced information| processing capabilities and labor cost advantages (Kumar tal, 2016), the transition away fiom human representatives in administering product and service offers seems predomi rately advantageous for firms. Despite AM's potential, seant research has examined how consumers evaluate AI systems in| relation o equivalent offerings delivered by humans ‘The increasingly pervasive use of Al raises the possibility that offers administered by Al agents may impact consumer response in novel ways as compared with offers administered by human agents. For example, Uber uses an Al machine feam: ing system f0 estimate travel elasticities and administer ride price offers (Newcomer 2017). Imagine a customer who pays foran User ride downtown but, forthe return rp, unexpectedly receives a price offer triple the orginal. Does administration of | sthis worsethan-oxposted offer by an AP agent, rather thas a human agent, have implications for purchase ikelibood, cus: tomer satisfaction, of intentions toward the use of Uber inthe furure? What if the return trip was unexpectedly much cheaper ‘than anticipated, resulting in a much beter-shanexpected offer for the consumer? The present research theorizes and demon- strates an interaction between the type of agent and eutcome expectation discrepancies, such that consumers respond les ron M Gave Anca Pofeor of Malsing and And 0 Reserch Protea of Maranng tan Coleg of Basra ad conan. Une fered, USA (ama AaronGarvey uty a) Tae Woo Ke Lec of Picken, Universi of Techn Syne, Ausra ema Taenon i ‘truss, An Ouluchex Profesor of Parken Uri of ino 1 Cseagn, USA ané Honorary Proestor of Marking, Univer dey Doses ora ncaa). ‘Supls by the Bech Ubrry 29 Jun 2022, 0751 (ST) negatively fo worse-than-expected price offers when tansmited by Al agents (compared with human agents) and more positively | to betersharrexpecied price offers when trnsmited by human agents (compared with AI agents). In addin, this interaction is furher moderated by anthropomorphic characterises of the A. ‘The more humanlike the Al's in terms ofits appearance or cognitive fictions the more it reduces the expectations discrepancy gap between human and AT agents, This moderator represents a static ‘manage input hat ean be used to manage customer susfcion| on the basis of the customer's price expectations, ‘Our theory and findings have key implications for fms enlisting both human and AF agents who administer outcomes that are disrepant fom expectations. These findings are rele= vant for price offers, in addition to other situations where con- sumers eam of unexpectedly negative outcomes along, dimensions ter than pres, such as cancellations, delays, nea ative evaluations, status changes, product defects, rejections, service fuilure, and stockouts. Our findings are also pertinent 1 instances where consumers receive unexpectedly positive ‘outcomes such as expedited deliveries, rebates, upgrades, sevice bundles, exclusive offers, loyalty rewards, and customer promotions. Managers can apply our findings to priontize (vs. postpone) human-to-AI role tanstions in situations where neg ative (vs. postive) discrepancies are more frequent and impact- ful. Moreover, our results suggest that even when a role transition is not holistically passed to an Al, the selective recruitment of an Al agent to disclose eertain discrepant infor- ‘mation can still Be advantageous. Fis that have already tan sitioned to eonsumerfacing Al agent, including the multitude of online and mobile applications tha use AL-based algorithns to eteate and administer offers, also stand to benefit fom our findings. Our research reveals that Al agents should be seloe- tively amthropomomphized (ie, depitod as either machinolike ‘vs humanlike) depending on whether an offer will be worse for better than expected Our research contributes to the literature in multiple ways Frist, we show that AT (9s. human) agents asymmetrically alter the effets of outcome expectation discrepancies on purchase at reengagement intentions, thereby broadening the keting contexts (eg, Huang and Rust 2018; Kim and Duhachek 2020; Longoni, Bonezzi, and Morewedge 2019: “Mende etal. 2019; Srinivasan and Abi 2021). Whereas the rowing Ineraure on technology in marketing has shown that ‘consumer engagement tends to decrease when on AL agent administers a transaction (eg., through a perceived lack of hhuman mental or emotional atrbutes on the part ofthe represen tative; Longoni, Bonezzi, and Morewedge 2019; see also Dietworst, Simmons, and Massey 2015), we reveal how transac- tions with Al ean either improve or undermine purchase tendon- cies and reengagement intentions depending on the valence of .05). No ‘ther regions of significance emerged at any higher levels of| anthropomorphism (for JN graph and supplemental table illus ‘wating interactions, see Web Appendix E, p. 14). That is, the positive effect of an AI (vs. human) agent on accepting a worse-than-expected offer emerges for individuals who are less likely to imbue the Al with humanlikeness, but does not hold for those individuals who increasingly attribute humanlike ‘qualities to technology. These results support Hi, Discussion ‘The results of Study 3a provide further evidence that worse-thanexpected offers are more likely to be aecepted ‘when administered by an AI (vs. a human) agent, in support Of H. Moreover, in suppor of Hs, individual pereptions of Al anthropemorphism alter this technology anthropomomphism of technology anthropomomphism ( ‘humanlike) result in responses simil ctl human, ‘These results have several important implications, Fist, our findings provide managers with a new, easly measurable vari able for segmentation when altempting to understand and predict how consumers will interact with their AI systems— individual tendeney toward technology. anthropomomphism Second, our results suggest that anthropomorphism of the Al through means other than innate consumer tendencies, such as physical embodiment or descriptions of the AI mind, could impact response to unexpected offers. Indeed, the results of Study 3a reveal that most individuals tend 10 view technology as less than humanlike (e., the scale mean for technology anthropomorphism was substantially and significantly below the midpoint; p <.001), suggesting that the default view held ‘by most consumers is that technology does not possess human like characteristics. In the next study, we examine how AL anthropomorphism can be manipulated, rather than measured, bby marketers to lead to a differential scoeptance of discrepant offers administered by AI. to those elicited by an ‘Study 3b: Anthropomorphism, Satisfaction, and Reengagement Study 3b expands on previous findings to examine how eus- tomer satisfaction and desire to maintain & relationship with the offering marketing agent varies depending on the level of humanikeness of the AP agent administering a discrepant ‘offer Customer satisfaction ts ertical for fms (Anderson, Engledow, and Becker 197%; Williams and Naumann 2011), ‘and evidence revealing how A anthropomorphism alters the impact of (dis)saisiction stemming from beter. oF warse- ‘Suppl by the Bish Library 29 Jun 2022, 0781 (BST) Garvey eta than-expected offers would provide valuable insights for man- agers of Al agents. Regarding relationship maintenance, build- ing « rapport between consumers and Al agents bas become an important issue for companies utilizing Al agents asa primary ‘contact point, with important implications for customer rete tion and relationship management (Huang and Rust 2018; see also Mende etal. 2019). As such, inthis study, we provide par- ticipants a choice to maintain (vs, replace) the relationship with the curent agent following a worse-than-expected offer. Consistent with our theory, we predicted that the effect of a betterthan-expected offer onthe tendeney to maintain versus replace the relationship would be stronger for a humanlike (vs. machineike) AL agent and that the opposite. patter ‘would result in the case of a worse-than-expected offer. To asses this prediction, we selected 2 novel pair of Al agents di fering only in ther perceived humanlikeness (ie, no diffe ences in perceived uncanniness), based on a comprehensive study of multiple robots drawn from the ABOT (hp: abotdatabaseinfo) database (development of Al stimuli detailed in Web Appendix ) Method Four hundred MTurk participants (Mg. =36.7, female =48%) completed this study for monetary compensation. Al patici- [pants were randomly assigned to one of the conditions in a 2 (agent typo: machinelike Al, humanlike Al)%2 (offer type Bete: than expeted, worse thin expected) beveen-ubjets esign. The overall procedure was similar to Study 2, which lized an Uber ridesharing context and explored customer sti faction as the dependent variable. Participants read a short passage explaining thatthe price offered by Uber was deter. mined by an AI agent (an initial $20 wip followed by a retum trip price offer of $30 in the worse-than-expected condition ‘and $10 in the betershan-expected condition). The images used f0 depict the two Al agents were dawn from the ABOT database (hitp:fabotdatsbase.info) and pretested regarding, their homanlikeness and uneanniness (pretest reported in the study "Results" subsection, addional deals in Web ‘Appendices F and G), Participants then answered tree items related to offer satis faction (identical measures to Study 1b; "satisfied," “apprecia- tive,” and “grateful” regarding the offer; 1="not at all,” 7= “very much"; the three items were averaged 10 create the index of satisfaction, a=.93). We next asked participants the extent (© which they wanted to interact with the same Uber agent in the future, which served as & measure of willingnesé| to reengage with the agent. Participants were given an apport nity to stay with the curent Uber agent in future interactions oF ‘discontinue the relationship with the current Urer agent and replace it with different agent in a binary choice paradigm, ‘The replacement agent stimulus was adopted fiom previous consumer research on AI (Mende et al. 2019; forthe exact replacement choice stimuli se Web Appendix F). Finally, par ticipants answered buckground questions (¢., gender) Results Protests, Separate protests supported the validity of our offer type and agent type manipulations, The pretest of agent type assessed the extent of perceived anthropomorphism (ie. hhumanlikeness) for each agent. Using the same three-item Fhumanlikeness (a-=,86) and thresitem uneanniness (a= 89) measures ffom our previous studies, a pretest of 100 MTurk respondents revealed that the humanlike (vs. machineike) agent was perceived to be higher in humanlikeness| (Moscistic™ 1.65, SD=.83; Myannits=247, SD 1.07; F(1, 98)=1805; p <0) but notin uncanniness perceptions (Moise 1-76, SD= 1.13; Mynine 207, SD 1.51; FL, 98)=135; p=25; additional pretest details and offer type pretest availabe in Web Appendix F). Offer sasfcton, The stsfition index was submited 1 a 2 (agent type: machinlike Al, humanlke.AD32 (offer type: tte than expected, worse than expected) ANOVA, The tain eet of offer type sas signieant(F(, 396)= 301.19, (p= 001, nf = 43) andthe main effect of agent was not signif cant (F(i, $96)=.03, p=.87, np < 01), Consistent with our ‘hoor, we also found a sgnfean interaction between agent and offer type FU, 396)= 1014, p=.002, 42.03) In the tertershan-expected offer condition, saisfation with the offer was higher inthe humanlike agent condition (M=5.79, SD=1.3) than inthe machineike agent condition (M= 5.29, SD=121; FC, 396)=5.2, p=.02, nf=.02). As preted, wwe found he ‘opposite patter inthe worse hanrexpeted ‘offer condition. Satsficion was higher in the machineike ‘agent condition (M=3.15, SD=1.84) than inthe umask agent condion (M=2.69, SD= 1.59; Fl, 396)=455, p= (03, nf = 02, sce Figure 4), This patern is consistent with our theo, providing direct suppor for Hs Choice of agents Oe ype (0= worse than expected, 1 =r than expected), agent (Ommachinclike, 1 shaman), and agent replacement decision (I= replace, O=rentn were const Coded and sited oa Binomial logic reson, We found Sipuean main els of offer ype Oe = 1.3, p sumed by a significant interaction between agent and offer spe (2 = 1241, p < 001). Further chi-square directional tests were ‘conducted fo assess the simple effects of agent type at each level of offer type. A worsehan-expeced offer led to a significantly higher proportion of replacement decisions in the humanlike ‘ageat condition (83%) than in the machinlike agent condition (63%) (one-sided test, p=.001; two-sided test, p=.002). Inthe betershar-expected offer condition, the proportion choosing replacement in the humanlike agent condition (28%) was lower than the machinelke agent condition (40%) (one-sided test, p= 04; tworsided test, p=.07). This patem is consistent with our theory and demonstrates that humanike AI is more (vs. les) likely to result in a continuation of the marketing relationship in the case ofan offer that is better (vs. worse) than expected. ‘Thus, we demonstrate that utilization of an appropriate AI ‘Supp by the Bish Lisrary 29 Jun 2022, 0751 (BST) Figure 4. Offer saisfacton asa function offer and Al agent type (study 35, Nowe ror brs 4S agent leads to stronger relationships with that agent, which is 4 critical factor influencing customer retention and relation= ship management, Discussion Study 3b demonstrated that consumer response to better than expected and worse-thanexpected offers ate systematically influenced by the administrating AL agent. Consistent with our theory, we found that a better-than-expected offer elicits seater sitisfction and reengagement when administered by a hhumantike (vs. machinelike) agent. These results ate particu larly important for companies utilizing Al a5 direct points of customer contact, with implications for the application and depiction of AL agents. General Discussion ‘The present research investigates how consumer responses to offers that are discrepant’ from expectations depend on ‘whether the administering marketing ageat is Al or & human, ur theoretical framework proposes that in the case of worsesthan-expected offers, consumers respond mare post tively when dealing with an AI agent versus a human oF more Thumanlike AI agen, inthe form of inreased purchase likeli hood, satisfaction,” and reengagement. In contrast, for berar-than-oxpected offers, consumers respond more positively toa human of more humanlike Al (vs. machinelike AI) agent. Moreover, we propose that this pattem emerges because Al (vs. humans) is infer to have weaker benevolent and selfish intentions, which in turn influences offer responses A series of five studies provides suppor for our theory across product and service contexts. In each study, an expectation- clserpant offer administered by an AI agent elicits different response relative to an identical ofer administered by a human ‘or more humanlike AT agent. Study la demonstrates in a product purchase contest that consumers are more likely to accept Journal of Marketing O(0) ‘worcthan-expected offers from an AL agent than human ageat. Study 1b examines the other side of our effect and demon- sires that consumers are moe likely to accept betterhan-expected offers from a human agent than an AL ‘agent. Study 2 reveals thatthe eect is driven by sronger inferred intendons on the part oF human versus AL agents; hati, songee ‘benevolent intentions inthe case of a btterthan-expected offer ‘and stronger selfish tention in the case ofa worse-shan-expected offer. These intentions, in turn alter consumer response. Study 3a reveals that an anthropomorphized (vs. a machete) AL agent leads 10 a lower acceptance of worseshanexpected offer. Filly, Study 3b examines how the physical embodiment of Al (Gc, a robot form) inluenees aniropomorphism, with implica tions for responses to diserepant offers. Administration of a better-thun-expected offer by a humanlike (vs. machinelike) Al results in higher satisfaction and reengagement intentions Together, these findings contribute to the literature examining hhuman~Al interactions and technology in marketing, in partic ular those areas examining preference, service, satisfiction, and anthrepomorphism, while identifying important implica: tions for consumers and marketers. Broad Implications of Differential Responses to Al Versus Human Agents ‘Our works reveal that fundamental differences exist bewen the ‘interpretation of negative and positive outcomes administered by ‘Al versus humen agents. The underlying mechanisms of perceived intentionality apply toa broad range of interactions beyond jus the poe discrepancies in our present work. Given this, other market ing and nonratketing contexts that deal wit the administration of negative and postive outcomes will key demonstrat similar pat tems of results as those proposed by jy, For example, consim- ers ate commonly faced with unexpected delays or cancellations (ea. fights, lodging stays, package deliveries), pesonalevalua- tions that differ ffom expectations (eg, work performance, ‘ret scores), discrepant financial offers (¢g, loan interest rates, ‘ret ine amounts), and unexpected school admissions and work: place hiring outcomes, in addition toa broad rae of ather eon texts im which AF pliys an inerasingly common and ‘evstomerfacing role. Puture research could also examine i the perceived faimess of an offer is affected by the type of agent FFor example, the positive or negative response 10 unexpected offers (ean unexpectedly low or high Joun interest rate) may polarize more extremely when administered by @ human (vs. AD) (or ial evidence in support ofthis prediction, see the supple ‘mental analysis of Study 2 ia Web Appendix D). Futute research could further examine the role of perceived faimess in explaining bow perceived intentions influence offer aceeptance When and How Marketing Managers Should Use Human Agents Versus Al Agents ‘Two pariculurly important marketing implications stem relly from the present research, both rooted in the ongoing ‘Suppl by the Brtsh Library 28 Jun 2022, 0751 (BST) Gorey et ab transition within firms from human agents to AT agents as mar- keting representatives. First, our findings give insight into how ‘his transition can be systematically and selectively managed 10 improve customer purchase tendencies, satissction, and reen- ‘gagement, Indeed, whereas extant works regarding customer- facing Al have revealed primarily main effects of Inumanso-Al transitions on consumer engagement—that i, cither humans or Al are generally superior in a given product or service domain (eg. Longoni, Bonezzi and Morewedge 2019; Mende ct al. 2019}—our work reveals that the nature of ‘the transaction (i, offers that are discrepant fom expectations) is important, In other words, our findings inform when and ‘where the transition should occur in product and service offer situations involving discrepant expectations. Managers could apply our findings to prioritize (vs. postpone) humar-o-Al role transitions in situations where betterthar-expected (vs. ‘worse-than-expected) offers are more frequent and impactful In addition, ou results suggest that even when a role transition isnot holistially passed to an Al, the selective recruitment ofan Al agent fo disclose worse-than-expected offers should sill be advantageous—that is, our results reveal that an AI “bad cop'thuman “good cop” approach to managing discrepant expectations should have beneficial outcomes for the fn Moreover, our research builds on brand crisis research that explores consumer respanse to algorithmic errors (Srinivasan and Abi 2021); however, our work eveals differences in per- cvived intentions, outside the context of algorithmic eros, related to whether an ofer is administered by a human or Al Second, and perhaps even more impactful, our work has implications for firms that have already transiioned 10 consumer-facing Al representatives, including, the multitude ‘of online and mobile applications that use AL-based algorithms | create and administer offers (eg., Myer's “6-second sale"; Greon 2017). In such postransition instances where AL is already customer facing, our research reveals that AI may be ‘depicted or otherwise framed as ether machinelike or human=| like, Our manipulations show that this can be accomplished through dliferences in the embodied appearance of the Al (ez. robotic form shown in Study 3b). Morsover, the resulis of Study 3a suggest that the majority of contemporary consumers consider Al to be inherently machineike (though Study 3a reveals that a minority of eonsuimers do already con~ sider AI to be moderately humantike) unless humanlike ‘marketers should unless specifically humanlike attributes are depicted. With this understanding, marketers should depict an AL agent as machine~ like when disclosing a worse-thanvexpected offer but should Sepict more humanlike attributes in the case of beter shan-expected offer, Ethical Implications The aforementioned recommendations carry ethical implica sions, as our findings reveal potential opportunites and con- coms for eustomer-fim relationships and consumer 1B wellbeing, On the one hand, our work reveals opportunities for firms to receive due, and perhaps otherwise overlooked, credit for bettersthan-expected offers and potentially other behaviors that are ostensibly benevolent. On the other hand, regarding threats to consuimer well-being, our work docs reveal tool through which marketers can increase the accep- tance of worse-than-expected offers made 10 a customer, hile also increasing intentions to engage withthe offending firm, In those instances where the worse-than-expected offer is objectively detrimental to consumers, the use ofthis approach ‘does raise ethical eoncems. By documenting these effets, we intend to benefit consumers and policy makers through stengt- ‘ning their understanding ofthe nature of differential consumer responses to discrepant outcomes. Importantly, se propose that ths ethical dilemma predomi nately emerges in situations where consumers are ostensibly harmed-—that is, where natural consumer resisanee to unex- pectodly high prices is bypassed by the selective use of Al However, some situations do exist where increased consumer parchase would not result in objective consumer harm, and there are even select contexts where consumers and firms ‘could jointly benefit fom ths effect. For example, consumers ‘often hold unrealistic expectations for product and service ‘offers due to selfsorving biases or lack of information. For ‘example, a customer could anchor on a historic promotional price (Lin and Chen 2017) or observe that another customer received a lower price offer without knowledge of that other customer's greater camed loyalty status (Mayser and Wangenhsim 2013), Such misperceptions or biased intepreta- tions could lead consumers o irationally abandon transactions that are objectively beneticil, In such cases, consumers and firms would benefit From shifing consumer tendencies toward preserving. the marketing relationship, thereby minimizing ‘switching cosis and maximizing consumer utility. Thus, the ‘current research indicates ways managers ean stengthen their ‘customer relationships through Al agents when these findings are applied ethically. “The results of our current studies reveal that most consumers ‘crrently tend 10 se¢ Al as not inherently possessed af enevo- lent or slfsh intentions when administering offers. However, ‘there i the possibility that some consumers infer that humans Wwithia companies are manipulating AL offers behind the soenes or otherwise have programmed the Al to indirectly ‘eamy out selfish human intentions. In such instances, the ‘observed increase in acceptance for” worse-than-expected offers from Al agents could be attenuated. Exploring whether ‘erin segments of consumers hold such skeptical belies, or whether similar beliefs may increase overtime as A adminsta- tion of offers becomes even more common, isa fruitful avenue for future research, ‘Anthropomorphism Theory ‘Our findings sso contribute to the literature on anthropomor- phism in product marketing and technology. Our research reveals that anthropomomphism of Al agents—that is, making ‘Suprliod bythe Bish Library 29 sun 2022, 07:81 (BST) 4 Alagents more humanlike—can actualy lead to decreased pre rence and increased disengagement. This isin contst to pre- vious research, which has shown predominately positive consequences of anthropomorphized products in the Form of increased consumer engagement and product liking (Aggarwal and McGill 2007). Wheres extant research suggests that humanlike Al i more comforting and trusted (Longon, Bonezzi, and Morewedge 2019; Waytz, Heafner, and Epley 2014), our findings reveal situations where humanlike AT may serve as aTability for maintaining the marketing eelationship, particularly in contexts of discrepant expectations. Future research should explore other psyeholinguistc and personality psychology factors incorporated into AT agent design (e altering the gender or perceived cultural origins ofthe Al) and resulting ekinges in consumer pereeption and reactions Beyond contextual effects due to Al design elements, our research also indicates that consumers individually vary in theic tendency to anthropomorphize Al agents, with subsequent implications for consumption. Our results reveal that individual uifferences in perceptions of technology anthropomorphism moderate offer acceptance in eases of worse-thar-expected ‘offers (Study 3a). Managers should consider integrating trait measures of technology anthropomorphisim when developing ‘consumer marketing segmentations relevant to Al interactions. Moreover, itis possible that other consumer-evel personality ‘mits wil in luence consumer-Al ineractions (ig five per- sonality factors)—a potential topic for futher researc. Associate Editor Juliet 2 Declaration of Conflicting Interests ‘The author(s) declared no potent! cont of interest with espect 0 the earch, authorship, andor publication of this ail Funding ‘The autos) eel no nancial support for he ese, autor andr publication ofthis acl References Asker, Jenifer, Kalen D. Vohs, and Cassie Moglner (2010), “Nonprofits Are Soon as Wan and For Profs as Content Fi Stereotypes Mat." Soumal of Consumer Research, 37 2), 24-3. Abel, Andrea E. and Bogdan Wojicke (2007), “Agency ond Communion from the Perspective of Self Versus Other,” Journal of Personality and Socal Procholog, 98 (3), 781-63 Aggaal Pankaj and An LMG 2007), Is That Car Sing at? Sele Conguty as 2 Basis foe Bvahating,Antropomanhid Prodi” Journal of Consumer Research, 34 (4), 468-79. Anderson, Ronald D., Jack L.Engledow, and Helmut Becker (1979), “Evaluating the Relaonships Among Attitude Toward Busnes, PreductSuisfction, Experience, std Seuch Elfor.” Journal of Marking Researol, U6 (3), 394-400. Journal of Marketing 0(0) Awad, Edmond, Sohan Dsoua, Richard Kim, Jonathan Schulz, ‘ose Hench, Azim Sharif tal, 2018), “The Moral Machine Experiment,” Notre, 863 (7729, 89-64 Boras, Alixandra, Emma £. Levine, Jonathan 2. Beman, and Deborah A. Small (2014), "Solis or Seldess? On the Signal ‘Value of Emotion in Alrite Behaviog,” Jounal of Personality ad Social Pryeholoy, 107 8), 393-413, Bausell, R. Barker and YwFang Li Q2002), Power Analisis or ‘Experimental Research: A Practical Guide for the Biological, Medical and Social Sconce, Cambridge, UK: Cambridge Univerity Press Bolton, Liss, and Anna S, Mattila (2015), “How Does Corporate Social Responsibility AMect Consumer Response to Service Failure in Buyer-Selee Relationships!” Journal of Retaing, 1 (0, 10-83, Brand, Myles (1984), fiend amd cing: Toward a Naualzed | ction Theor. Casidge, MA: MIT Press Braumaa, Michiel 6. (1987) atcarion, Plans, and Practical Reason, ‘Cambridge, MA: Harvard University Press (Chandler, Jesse and Norbert Schwarz (2010), “Use Does Not Wear Ragged the Fabric of Friendship: Thinking of Objects as Alive Makes People Less Willing 10 Replace Them,” Jounal of Consumer Psychology, 20 2), 138-45 ‘Chen, Rocky, Echo Wen Wan Peng, and Pie Levy (2017), “The Effect of Social Exchsion “on Consumer Preference for Anthropomerphized Brads,” Journal of Consumer Psychologs, 27 (1), 22-84, Darke, Pete R, Lauren Ashworth and Kelley J. Main (2010), "Great [Expectations and Broken Promises: Misleading Claims, Product Fae, Expectancy Disconfinnation and Consumer Diss.” owe ofhe Academy of Marketing Stone, 38 (3) 347-82. Davenport, Thomas, Abhijit Guha, Div Grewal, and Tina Brespott (2020), “How Atif Intelligence Will Change the Furure of Marketing.” Journal of te cademy of Marketing Science, 8 (1), 24-42, Dictorst Berkeley J, Joseph. Simmons, and Cade Massey 2015), “Algortim Avesion: People Erroneously Avoid Algortins Aer Sezing Them En" Jownal of Experimental Psychology: Gonorl. 1 (1). 14-26 Du, Stull, Chiabhun B. Bhutachary, and Sankar Sen (2007), “Reaping Relational Rewards from Coporte Socal Responsibility: The Role of Competitive Positioning.” International Journal of Rescareh in Marketing, 24 (3), 224-4 sera, Andre, Bret Kupre!, Roberto A. Novos, Justin Ko, Susin IM, Swett, Helen M. Blau, eta, (2017), “Dermatologi-Level Clasifeation of Skin Cancer with Deep Neva Nerwrks.” ‘Nature, $42 (1639), 115-18, Evangel, loannis and Sin MJ. Van Osseaer (2018), “Pons of (WigPaniy: Expectation Disconfimation from Commen Antibes in Consumer Choice" Journal of Marketing Research, 380. 13 Fisk, Raymond P, and Clifford. Young (1985), “Disconfimation of Eguity Expositions: Effect on Consumer Satisietion with Serices" NA ~ eldvamces in Consumer Reseurch, Vol. 12, Elvabeth . Hinehman and Moris 8. Holby, eds. Provo, UP: Association for Consumer Research, 340-48, ‘Suppos by the Beitsh Unrry 29 Jun 2022. 07:51 (ST) Garvey eto Gny, Kur (2012), “The Power of Good Intentions: Perceived Benevolence Soothes Pain, Increases Pleasure, and Improves “Taste” Socal Pchologial and Personality Seance, 3) 639-48, Giny, Heather M,, Kurt Gry, ad Daniel M. Wegnes 2007), “Dimensions of Ming Perception.” Sciene, 318 (S812), 619-18, Gna, Kur nd Daniel M, Wegner (2008), "The Sting of Itentonal Pain" PayeholoplealSetence, 19 (12), 1260-82. ‘Giny, Kort, Liane Yoong, nd Adam Waytz (2012), “Mind Perception Is the Essence of Moray.” Payehologiel Inguir, 23 2). lol ‘Green, Rick (2017), "Myer Launches the “6 Second Sale’ Campaign ‘ou YouTube via Clemenger BBDO, Melboure” Campaign Brief June 1), bps:/campaignrietcomnyer-anehes-he-5- secon Gite, Bema, Wouter van den Bos, and Eveline A. Crone (209), “Faimess Considesins:Ineasing Undersunding of Inentonaliy During Adolscenss,” Journal of Experinenal Cld Pokal, 104 4), 398-408, Harris, Karen, Austin Kimson, and Andrew Sehwedel (2018, “Labor 2030 The Collision of Demographics, Automation and Inesuiy,” Bain & Compas, research report (Febnary 7), pertain comfiighlsbee 2030checalisonctekmagrapic: utonatonandinegaly ayes, Andrew F (2017), Japextuaton 19 Meaton, Moderation, and Condional Process Anas: 4. Regresion-Based Approach Now York: Guilford Publications ilig, Benjamin & sbel Thins, fohanaa Hepp Sina A. Kein, ad Ingo Zetler 2015), “From Personality io Altusic Behavioe (and Back) Evidence ffom a. Double-Blind Dictator Game’ lournal of Research n Personality, 55 (Api), 46-50. Huang, Sauchi and Fangyan Chen (2019), "When Robots Come 10 ‘Our Rescue: Why Professions Service Robots Aren't Inspiting ‘nd Can Demotvte Consumers’ Pro-Social Behaviors” in Nt ‘Advances in Consumer Research Volume 47, Rajesh Bagchi, ‘Lauren Block, nd Leonard Lee, ed. Duluth, MN: Associaton fo Consumer Reseach, 93-98, Huang, Ming-Hui and Roland T. Rus (2018), “Avital Ineligenc in ‘Servie." Jounal of Service Research, 21 (2), 195-72 uagher, Clarctha, Lionsl Robert, Kristin Fry, and Adam Arroyo (2019), "Artificial Intelligence, Employes Engagement. Faimens, and Job Outcomes” ia Managing Techvolagy and Middie-ond LowStiled Employees. Bingley, UK: Emerald Publishing, 61-68 ‘im, TaeWoo and Adam Duhachek 2020), “Arif Intelgence and Persuasion: A Constual Level Account” Paychological Science, 31 (4). 364-80 im, Soca and Ann L. MeGill (2011), “Gaming with Me. Slot oF ‘Gumiag the Slot Machine? Power, Anthropomorphixm, and Risk Poreoptin,” Journal of Consumer Research, 38 (0, 5-107, im, Seo Young, Bemd H. Schmit and Nadia M, Thalmann (2019), “Han ja the Uncanny Valley: Anthropomorpizing Consumer Robots Incesses Ther Pereived Warmth but Decreases Liking Marketing Letters 30 (1), 1-12 Kumar, Vs Ashutosh Dini, Rajekar Raj G. Java, and ‘Mayubh Dass (2016), “Research Framework, Strategies, and is Applications of Inligent Agent Technologies (ATS) in Masketing,” Journal of the Academy of Marketing Science, 4 (2648, ‘Landebr Jan R. Ann L. MeGil, and Andreas Hermann (2011). "Is Got the Look: The Etfect of Friendly and Aggressive ‘Facil Bypressions on Product Liking and Sales" Journal of Marketing, 75), 132-46. Chien-Huang and Ming Chon (2017), "Follow Your Heat: How Is ‘Willingness to Pay Formed Under Multiple Anchors” Frome Paycholgs, 8 2269), 1-3. Longon, Chis, Andkea Bonezi, and Carey Morewedge (2019), Resitnce to Medical Anifeial Intelligence,” Journal of Consumer Research, 46 (8) 629-50. “MacDornsn, Kar F, Sanath K. Vasudevan, and Chin-Chang Ho (@008), “Does Japan Really Have Robot Mania? Comparing Attitudes by Implicit and Explicit Measures." AT & Sacets, 23, 495-510. Malle, Bestam F., Louis J. Moses, and Dare A. Baldwin (2001), “Inroduetion: The Significance of Inentonsiy Inteatons and Intentanalty: Foundations of Social Cognition, 4), 126, Mayser, Sabine and Florian yon Wangenheim (2013), “Pecived Faimess of Differential Castomer Trestment: Consumer Understanding of DistnbutveJatice Really Mares.” Journal of Service Research, 16 (1), 99-113. Mende, Marin, Maura L. Scott, Jenny van Doom, Dhniv Grewal, and ‘Hana Stanks (2019), "Service Robots Rising: How Humanoid Robots Infuence Service Experiences and Elicit Compensatory Consumer Responses" Jounal of Marketing Research, $6 (4), 535-56, Morewedge, Carey. K. (2008). "Negativity Bias in Atibution of Exteral Agen.” Jounal of Experinental Pecholoy: General, 138 (2, 535-48, Mori, Masthir, Kal F. NeDarman, and Noe Kageki(2012),“The ‘Uncenny Valley,” IEEE Robotic & Ausomation Magazine, 19 2), g-100, Newcomer, Erie (2017), "Uber Stats Charging What Thinks You're Willing « Pay,” Bloomberg Nevs, (May 19), hips-ivwn ‘ombrg somfewnartle/2017-05-19kherfaurenny- ray on-proditing-how-much-you-e-willing 0-70) ‘liver Richard L(1980)."A Cognitive Model ofthe Antecedents and Consequences of Siisfiction Decisions,” Journal of Marketing Research, 17 (4), 460-69 Oliver Richard L. P.V. Sundar Balakrishnan, and Bre Bary (1994), "Outcome Satifiction in Negotiation: A Test of Expectaney Disconfimation,” Orgonzaronal Behavior and Human Decision Processes, 0 (2), 252-15, Oliver, Richard L. and Wayne S, DeSatbo (1988), "Response Determinants in Satstction Judgment,” Journal uf Consumer Research, V4 (8), 495-507, Pollak, Jordan B (2006), “Mindless Ieligence,” BEE Ineligen ‘Sate, 21 (3), 50-6. Radke, Sina, Bara Giro, and Elen RA. de Bj (2012), “There's Somsting About 2 Fir Split: Intesonality Moderaes Context Based Fsimess Corsets in Social Decision-Making” PLoS One, 70. Li ‘Suppl by the Bish Lirary 28 Jun 2022. 07:51 (BST) 6 Rand, David G., George E, Newnan, and Owen M. Warzacher| (2015), “Social Context and the Dynaries of Coopertive Choice” Journal of Behavioral Decision Making 28 (2), 138-66. Rezs, Paul and Carol Nemeroff (2002) “Symptistic. Magical ‘Thinking: The Contagion and Similarity “Hearse.” in Hewisne and Biazes: The Pxichology of Imutive Judgment ‘Thoms Gilevich, Dale Geifin, and. Daniel Klien, el New York: Cambede Univesity Press, 201-16. Russell, Start J. and Peter Norvig 2010), Arial Ineligence: ‘Modern Approach, Hoboken, Nl Pearson Education. Sanfey, Alan G. James K. Rillng, Jesion A Aronson, Leigh TE Nystom, and Jonathan D. Cohen (2003), “The Neural Basis of eonomie Decision-Making in the Ultimatum Game.” Science, 300 (5626), 1755-58, Spreng, Rickard A.,Scox B. McKenzie, und Rican W. Olshavky 11996), "A. Reexamination of the Deteminants of Consumer Susan," Journal of Marketing, 60 (3), 15-32. Seinvasan, Raj and Glen Saal Abi (2021). "When Algorithms Fail: Consumers Responses to Brand Harm Crises Cased by Algorithm rors" Journal af Marketing, 85 (5), 73-91 Slinan, Ram” (1998), “Expectations and Faimess in a Modified Ursa Gare" Journal of Eeononde Palo, 17 (5), S31-84 ‘Thaler, Richard H. (1988), “Anomalies: The Ultimatum Game,” Jal of Beonomie Parspectves, 2 (2), 195-206. “Tis, Michael, Vikus Mita, and Willian, Ross J. (2004), “The Role of Atibutons in Customer Satisfiction: A Reexannation.” owrnal of Consumer Research 31 (2). 416-83, ‘Supp Journal of Marketing 0(0) “Tunot, Karen (2016), “Mest "Ros the Newly Hired Leysl Robo." ‘The Washington Post (May. 16), bps:/www washingtonpos. comneweinnovatonswp/201605/16imeet-rosshe-newl-hited- lepa-obo ayer, Adan on Casiopp, and Nichols Epley (2010), "Who Sees Hi? The Stability and Importance of Individual Differences in Antropomorphisn,” Perspectives on Psychologica Science 5 (3), 219-22 Wytz, Adam, Joy Heater, and Nicholas Epley (2014), The Ming in the Machiae: Andropomorphism Ineresses Tustin. an ‘Autonomous Vehiele” Jounual of Experimental Socal Peycholgy, 82 May), 113-17 ‘Wiliams, Paul and Ear Naumann (2011), "Customer Saison end Business Perfomance: A FimLevel Analysis” Jounal of Services Marketing, 25 (1), 20-32 ‘wine, Jochen Pal G, Paterson, Werner H. Kunz, Thorsten Graber, ‘Vink Nhat L, Sefnic Paluch ea. 2018), "Brave New World: Service Robots i he Frontline,” Jounal of Svice Management. 29 (5), 907-31 \Wojeizke, Bogdan, Andres B, Abele, and Wieslaw Bara (2009), “Two itetsons of interpersonal Aditdes: Liking Depends on ‘Connon, Respect Depends om Agen." Burgpean Jounal of ‘Social Pcholog, 39 (6), 973-90, Zlowwski, Jakub, Diane Provdfot, Kumar Yogeesvaran, and ‘Christoph Barteck (2015), “Ambropomorphism: Oppornites ‘and Challenges in Hman-Robot Interetion,” Inernational ournal of Saal Robotics, 7 (3), 34-60. bythe Bltsh Library 29 Jun 2022. 0751 (BST)

You might also like