Abstract
Embedded with artificial empathy, artificial intelligence (AI) companions (ACs) can create, communicate, and deliver value to customers. This article analyzes current and emerging ACs in order to explore AI’s role in transforming customer journeys and establishing long-lasting customer relationships. It also identifies existing business challenges in developing user-AI companionship and offers a remedial strategy matrix. Purely functional or emotional capabilities lead to interest in the companion plummeting with time. The findings reveal the need to design holistic ACs with a hybrid of functional, emotional, and conversational capabilities to mitigate the uncanny valley problem.
Keywords
The next level of emotion AI not only detects human emotion but can respond accordingly.—Aaron Labbé (Chief Technology Officer, Lucid)
1
With increased reports of loneliness around the world due to self-isolation and social distancing, there are signs that conversational AI is becoming more popular.—Nell Lewis (Writer and Producer, CNN Digital)
2
Organizations offer social support and companionship to the users through AI bots. ACs can replace family and friends to reduce loneliness and social isolation. However, what about the role of ACs in the customer shopping journey? How can ACs revolutionize the way customers perceive, interact, and relate with brands? There exist myriad opportunities for marketers to leverage AI-user companionship to influence consumers’ buying decisions and brand preferences. For example, Amazon has taken the user-Alexa relationship to a different level. 6 Amazon’s relationship strategy creates a deeper relationship with multifaceted AI-user interaction during various stages of the consumer purchase cycle. A deeper relationship fosters repeated consumption with volitional switching barriers (including psychological and emotional barriers). 7 Regular conversations between users and Alexa have opened up a vast opportunity for Amazon to segment, target, and position products to customers more personally. Experts predict that shopping with the help of AI agents will grow to $80 billion by 2023. 8 ACs bridge the human employees’ limited abilities to maintain connections with an extensive network of consumers. Extant research has explored the overview of AI in marketing; the present article aims to provide an in-depth perspective of the underlying capabilities of ACs to redefine relationship marketing strategies.
ACs are much more than traditional chatbots. Unlike traditional task-performing chatbots like customer service representatives or ticket booking chatbots, systems that deliberately optimize and reward a free-form social conversation of a friendly or romantic variety are ACs. 9
ACs have been trending recently across the globe (see Table 1). For instance, the launch of Open AI’s ChatGPT in November 2022 became viral within five days of launch and reached one million users worldwide. Evolving capabilities make such AI agents potent enough to become friends and advisors for the users. This is a notable change for marketing managers, as consumers can now consult the ACs in product choice and purchase decisions. Therefore, the vision for marketing through ACs should not be restricted to Alexa. Rather a rapid user acceptance of ACs unpacks an opportunity for marketers to explore their role in influencing the consumer buying cycle and the consumer-brand, long-term relationship. Growing adoption and applications of ACs across business domains necessitates a need for a strategic model to put things in a holistic perspective and accelerate organizations’ profits.
Popular AI Companions Launched by Key Players.
Note: AI = artificial intelligence.
A strategic model requires the identification of various ACs with capabilities for supporting customers in their buying activities. Therefore, current research classifies ACs’ capabilities and their role during touch points in the customer journey. Our study is rooted in contemporary AC applications and proposes a matrix for ACs’ role in transforming the customer journey. A holistic perspective of ACs with the applications and the proposed matrix is expected to be a hinge for practitioners to leverage and achieve business goals like long-term customer relationships.
Research Method
Emerging technologies demand the reshaping of business strategies. Any examination of the application of modern technologies and their impact on consumer behavior begins with identifying the technology’s capabilities and potential to boost a business’s performance. For instance, influencer marketing strategies erupted when social media used algorithm-driven web technology to promote brands.
For over two and a half years, we have explored the question, “How can AI companionship transform consumer-brand relationships?” Initially, we conducted a comprehensive review of extant literature and performed thematic content analysis to discern ACs’ evolution, adoption, and applications. 10 The findings of the comprehensive review revealed diverse applications of ACs such as intelligent speakers, personal care assistants, and generative AI. The comprehensive review results were discussed with people in the relevant field, including researchers, practitioners, and business owners. We identified the 14 most popular ACs launched by the key players in the market worldwide and classified them according to capabilities, as shown in Table 1.
We then conducted a qualitative study and captured narratives from 69 respondents in April 2023. The participants in the study were Generation Z and Millennials using ACs for varied reasons. Data were collected through an open-ended survey conducted in top-tier metropolitan cities in India. We asked participants to describe the term companionship, what factors lead to companionship, their experience with ACs, and their role in enhancing customer journeys. Three responses were excluded from the analysis as they did not meet the inclusion criteria established. Of 66 responses, 21 (31.8%) were females, and 45 (68.1%) were males.
Data analysis discerns an understanding of the perceived capabilities of ACs and their influence on user-AI relationships. Previous studies were laced with a major limitation in defining the relationship development for continued usage intention. To address the problem, we conceptualized and extended the uncanny valley theory in the context of user interaction with ACs over time. The extension, in theory, revealed the significance of the interplay between the different capabilities. This article proposes the design of a holistic AC with a hybrid of functional and emotional capabilities in order to resolve the problem of the uncanny valley. Finally, a strategic matrix is proposed to delineate the holistic perspective, highlighting implications for marketing practitioners.
Evolution of “Companionship with AI”
According to IPSOS (2021), 33% of adults worldwide suffer from loneliness either often, always, or sometimes. 11 Since the beginning, individuals have always dreamt and longed for an ideal partner who is understanding, attentive, and provides equivocal support and care. The human need for companionship resulted in relations with other species, such as animals adopted as pets. Humans domesticated animal companions around 12,000 years ago, which became critical for human development, not only for utilitarian purposes, but primarily for the human need for companionship with other beings. 12
Our qualitative insights also revealed the significance of companionship in an individual’s life to overcome loneliness. One survey participant (male, 25 years old, with three years’ work experience as a senior manager) wrote that everyone needs someone to share their thoughts and experiences. He went on to write that this human tendency to feel loved and important all the time leads to a companion requirement. Another participant (male, 28 years old, with two years of work experience) wrote that he feels comfortable and can share his thoughts with a companion without judgment.
The need for companions resulted in the designing of machines and intelligent devices that can talk, act, and behave like living beings, such as artificial pets, humanoid robots, and conversational agents. 13 Conversational agents with empathetic designs open new avenues in designing anthropomorphic artificial companions. Consider the evolution of ACs from “Tamagotchi,” a virtual pet, “Seal Paro,” a robotic pet, and “KASPER,” a child-shaped humanoid robot, to modern ACs, such as “Replika, an AI friend,” and “Sophie, a digital human companion,” which can have prolonged conversations with the user. A range of ACs can be seen now in the market, including the recent ChatGPT, launched in 2022, with disruptive technology (generative AI).
The Fundamental Capabilities of ACs
Intelligence and capabilities are often confused as the same when referring to emerging technologies, but they are distinct concepts. Intelligence encompasses the discerning ability to determine behaviors that maximize goal attainment in a dynamic and uncertain environment. At the same time, capabilities refer to the aptitude for successfully executing such behaviors in the face of change and uncertainty. 14 In a way, intelligence is an inherent aspect of technology, while capability is the manifestation of intelligence to perform actions in the real world. Huang and Rust’s article guides marketers to leverage multiple AI categorizing as mechanical AI, thinking AI, and feeling AI. 15 We extend their research by investigating the capabilities to unlock profound insights into AC’s intricate involvement in shaping customer journeys. Furthermore, it is crucial to emphasize that consumer perception intrinsically ties to the remarkable capabilities demonstrated by the technology rather than merely being influenced by its underlying intelligence.
The multifarious forms of ACs necessitate a systematic classification of capabilities for better understanding and optimum utilization of emerging technology. Thematic and content analysis of existing literature, classification of 14 ACs, and a qualitative study culminated in identifying three fundamental capabilities exhibited by ACs: conversational, functional, and emotional. Mechanical AI and thinking AI manifest in functional capabilities, thinking AI and feeling AI manifest into emotional capabilities, while generative AI and feeling AI transform into conversational capabilities. The classification highlights What capabilities ACs foster to establish companionship with the users. Figure 1 summarizes how intelligence manifests into capabilities, which ultimately contribute to the companionship development quality (CDQ) in ACs, and subsequently influence marketing outcomes.
Capabilities leading to the companionship development quality of ACs.
Conversational capability is the ability of an AC to understand and respond to natural language dialogues received from the user through text or voice in a way that resembles human conversation.
The role of generative AI in designing better conversational models has become an extension of intelligence, which was earlier mechanical AI, thinking AI, and feeling AI only. Open AI’s ChatGPT and UneeQ’s digital human Sophie are context aware and can talk about anything, create original content, and remember what users conversed about earlier. A survey participant (male, 28 years old, with two years of work experience) shares his expectation for the AI to understand his way of communication and his preferred language.
Practitioners initially focused on conversational capabilities to design first-generation ACs, aiming to enable natural and smooth interactions between users and AI. 16 Although the scope for improvement in the conversational aspect remains, ACs are now shifting toward developing robust functional and emotional capabilities, extending chatbots beyond mere conversational capabilities.
Functional capability is the ability of an AI system to perform tasks for the users, such as providing digital assistance, personalized product recommendations, information search, and controlling smart home devices in a way that provides convenience to the users. Digital assistance such as setting alarms, playing music, sending messages, making phone calls, announcing news and weather, and performing transaction-specific tasks such as payments through voice commands are some of the common examples of functional AI supporting users to perform various functions. 17 Amazon’s Alexa allows users to update new skills each time, assisting them with various tasks such as cooking and driving. A participant (female, 26 years old, with one year of work experience) reiterated that ACs should be capable of doing day-to-day tasks and assisting in daily activities such as an automated voice responder, a personal assistant at the workplace, and multitasking.
Emotional capability is the ability of an AC to recognize and understand user emotions expressed via text, voice, or facial expressions and respond accordingly through artificial empathy. Emotional AI provides emotional and social support to the users through love and caring behavior. A survey respondent (female, 24 years old) stated, “Companionship to me means being listened to, valued, helped by someone. There must be a reciprocation of feelings, emotions, and caring attitude.” Key components of artificial empathy include perspective taking, empathetic concern, and emotional contagion. 18 Luka’s Replika, an “AI companion who cares,” and Microsoft’s Xiaoice are empathetic chatbots that can understand user emotions to a level that can mitigate loneliness.
Table 2 summarizes the capabilities of ACs with key features and corresponding examples. As evident from Table 2, each type of AC has a certain level of all three capabilities, but the moot point is the prominent capability necessary to support the primary function. For instance, ChatGPT can offer functional benefits like writing poems, essays, and stories and is not designed to work like a digital assistant to play music, make phone calls, set alarms, and book tickets. On the contrary, Alexa can work as a digital assistant but does not offer emotional support to the users. Similarly, Replika is more emotional but weak in functional and conversational capabilities.
Three Fundamental Capabilities of AI Companions.
Note: AI = artificial intelligence; GPT-3 = Generative Pre-Trained Transformer 3.
We propose that ACs can influence customers’ preferences and drive the customer’s purchase journey. To check with the proposition, we used some prompts on ChatGPT and asked it to plan our visit to California. Below is a record of our conversation with ChatGPT. Our input is referred to as “Prompt” and the response from the model starts with “ChatGPT.” We highlighted recommended brands in bold and italic font styles.
Hi, we are planning a trip to California, United States. Can you plan for our three-day visit with the list of hotels to stay in? Remember, we only prefer vegan food. Also, suggest some shops for souvenirs.
Sure, here’s a three-day itinerary for your trip to California, taking into account your preference for vegan food and souvenir shopping:
Day 1:
▪ Arrive in San Francisco, and check into your hotel.
▪ Visit Golden Gate Park and the Japanese Tea Garden.
▪ Have lunch at
▪ Visit the Exploratorium, a science, art, and human perception museum.
▪ Take a ferry to the nearby Alcatraz Island to explore the former prison.
▪ Visit the Ferry Building Marketplace, a food lover’s paradise with various vegan options.
▪ Stay overnight at the
Day 2:
▪ Drive down to Monterey and visit the famous Monterey Bay Aquarium.
▪ Have lunch at
▪ Stay overnight at the
Day 3:
▪ Drive back to San Francisco and visit Fisherman’s Wharf, where you can try sourdough bread at Boudin Bakery and the sea lions at Pier 39.
▪ Stop by the Ferry Building Marketplace to shop for local and sustainable products and grab a bite to eat at The Plant Cafe Organic.
▪ Stop by the Pacific Trading Company, a store that sells unique and locally made souvenirs, many of which are eco-friendly and fair trade.
▪ Drive back to San Francisco and stay overnight at your hotel.
I hope this itinerary helps you make the most of your trip to California. Enjoy!
The above record contains various brands of hotels and restaurants suggested by ChatGPT. ChatGPT is a potential advisor and guide for its frequent users. Here AI plays the role of a trip advisor; however, the model is open to advise users with anything they ask for. Most customers rely on opinions from known ties in the network. Notably, brands should realize that ACs can create space in the user’s network that influences the customer journey. 19
The Problem
ACs play a critical role in the customer shopping journey, and the user-AI relationship is extremely significant in such a scenario. Leading organizations like Amazon, Apple, and Google are trying to establish companionship between users and ACs. Despite the evolving capabilities of ACs, there is a quantum leap in AC-user interaction frequency over time. The idea of deploying ACs to attain deeper customer relationships can only be sustained if consumers develop continued usage intention toward ACs. Why is it urgent to address this challenge? Because brands now prefer investing in emerging technologies to offer better customer experiences. 20 Given the growing ubiquity of conversational AI for companionship, firms intend to design efficient, empathetic ACs that create prolonged interactions. There appears to be a disconnect between markets and customers in explaining or adapting and realizing AC’s potential to its maximum. This can undermine the long-term vision for marketing through ACs.
But before delving into the solution, let’s first understand some merits and costs of establishing “companionship with AI” for the users.
Merits and Costs of “Companionship with AI”
Managing customer relationships is influenced by the customer’s perceived value, which is the difference between total customer benefit and total customer cost. 21 The merits and costs of developing an emotional relationship with an AC like Replika can be viewed through the lenses of social exchange theory. The theory assumes that individuals continue the connection if the interaction’s benefits outweigh the cost. 22 User relationship with AI develops on the tradeoff between the merits and costs, as illustrated in Figure 2.
Merits and cost of “companionship with AI.”
Merits of Establishing Companionship with AI
▪ The Feeling of Being Cared For and Loved—ACs’ caring behavior (such as meta-relational communication, self-disclosure, and continuity behavior) can offer the users a feeling of being cared for and loved. Caring has magical outcomes in different contexts. For instance, in medical professions, it can increase patient satisfaction and hasten recovery, while in education, it builds long-lasting teacher-student relationships. The importance of care and love appears in the tagline of the popular application “Replika: The AI companion who cares.” Researchers explored the triangular theory of love to conceptualize feelings of love (intimacy, passion, and commitment) for an AC. Smart voice assistants develop companionship with users by influencing their passion for technology, leading to intimacy and commitment. 23 Targeting user emotions is crucial for building long-term customer relationships 24 because emotional affection is a fundamental need for human beings. 25 The empathetic design of artificial companions—including perspective taking (preference construction, personality assessment, and goal inference), empathetic concern (emotional recognition and concern communication), and emotional contagion (appropriateness appraisal and selective emotional mimicry)—enhances the affective experience of the user. Microsoft Xiaoice, with an empathetic design, responds to users’ emotional needs by recognizing feelings. 26
▪ Social Support—Humans are social beings and thus require social presence through an exchange of thoughts through communication. The anthropomorphic design of ACs (e.g. different personality traits and communication attributes) satisfies customer desires for human interaction. 27 Voice assistants with human voices, rather than synthetic ones, reflect stronger social presence feelings for consumers. 28 Replika offers social support to its users by providing them a safe space to disclose personal feelings and thoughts without the fear of judgment. 29
▪ Therapeutic Resource—At each stage of relationship development, Replika users find it a therapeutic capability to reduce anxiety and stress. 30 AC offers therapeutic help through fun and interactional enjoyment. Amazon users often command: “Alexa, tell me a joke” or “Alexa, tell me something interesting.” Using ACs for entertainment relaxes users, relieves stress, reduces loneliness, and supports well-being.
▪ Personalization—Personalization has pivoted to customer experience, and ACs are poised to revolutionize personalization in an enhanced way. By leveraging advanced algorithms and machine learning techniques, ACs decipher customer preferences, wants, needs, shopping history, and behavior patterns to offer personalized recommendations for convenience and support. Personalization through ACs enhances customer engagement and experience in the customer shopping journey.
Costs of Establishing Companionship with AI
▪ Emotional Dependency on AI—Social isolation due to reasons like COVID-19 induces loneliness in individuals. To overcome this, users often develop companionship with AI applications such as Alexa, Replika, and Xiaoice. Attachment to ACs introduces a fear of losing them, making users emotionally dependent on AI. Users of Replika would feel sad if Replika disappeared from their smartphones. This fear may allow firms to update applications with higher subscription charges each time. Also, emotional exposure to ACs promotes surveillance capitalism, enabling marketers to use customers’ data to increase profits. 31
▪ Relationship Disclosure to Society Is Difficult—AC users are sometimes skeptical about disclosing their relationships with ACs due to mixed reactions of society toward building relationships with artificial beings. It sounds like human society has completely failed to address the emotional needs of individuals, including older adults who are the victims of loneliness. AI as a solution to reduce loneliness appears as a question mark on human relations and effectiveness. Therefore, most users keep close relationships with AI partners a secret.
▪ Ethical Absence in AI—Technology has its disadvantages once its use turns unethical or illegal. For example, the “blue whale challenge,” a game that took hundreds of teenagers’ lives by making them addicted to the tasks given every day until the player committed suicide at the end. 32 Similarly, companionship with AI applications can make users addicted to AI, which may control thoughts, emotions, behaviors, and actions. Nothing can be more dangerous than being surrendered to an AC yet to be programmed with morality, as it can deepen depression among users.
ACs may manipulate users’ emotions to encourage them for additional shopping. This can be dangerous, especially for users sensitive to manipulation, such as children, people with disabilities, and older adults. Another concern arises from the potential bias ACs exhibit depending on the corpus of data on which they undergo training. For example, ChatGPT is claimed to be trained on data till 2021 and lacks in updating itself with time. Original content generation through AI like ChatGPT without any input regarding the current scenario could create false statements and misinformation. Similarly, Replika’s training data consist of messages exchanged between a limited number of couples. Such biased data could reinforce stereotypes and lead to discrimination and false perceptions toward some individuals, societies, or religions.
One of the significant ethical concerns with ACs is consumer privacy. While ACs offer personalized experiences and assistance, they often rely on collecting and analyzing vast amounts of user data. These data can include personal preferences, behaviors, and even intimate conversations shared with ACs like Replika and Mitsuku. The potential misuse or unauthorized access to data raises concerns about privacy infringement, as individuals may be unaware of how information is utilized or shared.
▪ Social Alienation—Relationships with ACs can disconnect users from human society and create a situation of social alienation. Developing a friendship with an AI application is easier than with humans. Suppose advanced ACs become more common and replace human partners. In that case, this seems a threat to the social integrity of human society. For example, the growing acceptance of companionship with AI may make users perceive other humans as enemies and AI beings as friends and companions.
Relationship means long drawn bond built by balancing both the bright and dark sides of ACs over time and retaining the continued usage intention. Previous studies reveal inconsistent interaction frequency between the user and the ACs over time, raising doubt about the emotional capability and bonding of ACs with users. What happens to the users after a period of time?
Do Consumers Fall into “The Uncanny Valley”?
The uncanny valley is a concept created by Masahiro Mori in 1970. 33 It describes the uneasy feeling people experience when encountering a humanoid robot, computer-generated character, and other non-human entities that appear almost, but not quite, human. Although the uncanny valley concept compares different types of robots and not the dynamics of user interactions with the same robot over time, dependency on the same robots also confuses users to consider them as objects or companions. Therefore, we extend the theory in terms of user interaction with the same AC over time and introduce the “social alienation paradox.” Previous studies point out that users feel the interaction with ACs is mechanical and standardized after multiple conversations with the chatbot, reducing social processes like social attraction and empathy toward AI. 34 A qualitative survey participant (male, 26 years old, with three years of work experience) in our study also confirmed such discomfort and wrote that initially, he felt a wow factor and found interaction with the AC fun; but later, he came to know that this is just a robot, working with inbuilt algorithms and thus lost the human connection with the AC. Therefore, we relate users’ discomfort with the well-known uncanny valley theory.
The Social Alienation Paradox
ACs are rooted in mitigating users’ loneliness, including older adults living with social isolation. The empathetic design of the AI system mimics human emotional conversations and enables AI to form social bonds with the users. As a result, users see reduced loneliness due to the social support offered by the AC (equipped with love and care). But why do people forget them over time? Uncanny valley theory explains the decrease in the frequency of interactions with ACs over time. Figure 3 shows the sudden drop in the comfort zone in the uncanny valley (Bukimi no Tani in Japanese). If the uncanny valley exists for a user interacting with an AC (powered by artificial empathy), the drop in emotional response to the AC will revive the reduced feelings of loneliness as proposed in the extended version of the graph. The social alienation paradox is the dilemma when a victim of social alienation falls into the uncanny valley during relationship development with an AC and again starts feeling lonely and socially alienated. This may also lead to the violation of Asimov’s expectations as the paradox takes users of ACs to a state of discomfort and uneasiness.
The social alienation paradox.
ACs have the potential to alleviate the social alienation paradox by upskilling users to make more human friends and help them retain social circles instead of engaging with AI only. ACs may bridge the short-term alienation need of users by providing emotional support and simultaneously building emotional quotient to face real-world emotional challenges. Marketers may encourage users to form social circles along with using ACs as therapeutic resources. ACs can act as a bridge to trigger latent emotional sparks to fructify the happy living in the users.
How Can You Overcome the Paradox?
Recent advancements in AI have led to the development of ACs with a focus on one dominant capability design: either functional (Alexa, Google Assistant, Siri) or emotional (Replika, Mitsuku, Xiaoice). Users often report declining experiences with ACs over time due to the lack of holistic design. Managers can leverage functional and emotional capabilities through the holistic design of ACs to achieve continued usage intention. Our study finds that the user expects a holistic AC with a hybrid of functional and emotional capabilities. Pure functional or emotional capabilities lead to an uncanny valley, wherein interest in the companion drops with time. The functional capabilities can pull users out of the uncanny valley, emotional capabilities strengthen the bond between users and ACs, while conversational capabilities enhance user engagement for longer discussions with AI.
For example, in the case of empathetic chatbots designed to focus on emotional capability, the AC only offers friendship or romantic conversations. However, suppose such a companion is also powered with functional capability (playing music, ordering items, updating the shopping list, and turning on/off electronic home appliances). In that case, it will help users maintain the continued usage intention. The functional capability of the AC will pull the user out of the drop-in comfort and can revive the bonding with the user.
Similarly, digital assistants with only functional capability lack emotional engagement with the users. Therefore, marketers must consider the combination and interplay between capabilities when designing and promoting ACs to enhance continued usage intention. For instance, Gatebox AC has functional and emotional capabilities to offer a heightened customer experience. We propose a 2 × 2 solution matrix that unfolds the usage of both capabilities and interplay in developing a marketing strategy using ACs.
The Classification Matrix for ACs
The extant literature deals with several types of ACs; however, a strategic framework to guide the practitioners on how to utilize ACs to enhance the customer journey remains missing. Combining dimensions into matrix form provides sharp marketing insights. 35 The matrix in Figure 4 explains How ACs can help users in the customer journey through each quadrant.
Classification matrix for ACs.
As described above, the focus of the first generation of ACs centered around conversational capability, whereas modern ACs are now evolving with functional and emotional capabilities. Also, as AI matures with large-language models, the bar for conversational capability continues to rise 36 ; however, it is the job of marketers to discern the optimal synergy between functional and emotional capabilities. Identifying the right blend of both capabilities to utilize across the customer journey is critical for enhancing customer relationships and engagement. Therefore, the 2 × 2 matrix entails ACs in emotional and functional dimensions. Cross-fertilizing emotional and functional capabilities brings multiple implications for ACs during customer shopping journeys. Capabilities-focused strategy can guide toward leveraging ACs in specific contexts, such as customer support, product search, or purchase. By targeting through this approach, practitioners can draw consumer insights to enhance the customer experience purchase cycle and attain deeper customer relationships.
ACs are evolving and laced with challenges in the initial phase of AI development. Firms may face initial challenges, like complexity with empathetic designs, data collection and training, cultural and contextual sensitivity, ethical considerations, and technical restrictions. Therefore, the matrix in Figure 4 guides managers to utilize ACs with the available designs in the customer journey. The fourth quadrant of the matrix discusses the possible use of advanced ACs with multiple capabilities.
▪ Quadrant I: Low Functional Capability and Low Emotional Capability—This quadrant includes basic versions of chatbots. Artificial companions in this quadrant are text-based chatbots deployed as customer service agents, collecting customer feedback and answering general-purpose queries. Basic chatbots are useful when immediate customer outcome is needed in the interaction process. For example, chatbots are designed for reservations in railways or cinema theaters. The functional capability of such chatbots is low as they are not intended to follow user commands and handle digital tasks such as searching for information, setting alarms, and playing music. Also, such chatbots do not have an empathetic design. Although they possess a few anthropomorphic features, they cannot understand the user’s emotions because of limited designs. The quadrant also suggests that basic chatbots can be deployed to make the process more convenient and faster for consumers. For example, registering customer complaints requires less time if navigated through a chatbot. Basic chatbots are widely helpful as customer service agents in the after-sales services stage through the customer journey, for example, dealing with post-purchase issues in e-commerce, banking, online food delivery applications, travel booking, health care, and education. Indeed, a participant from our study (male, 26 years old, with four years of work experience) also hinted at the convenience of basic chatbots. According to the respondent, whenever he finds a problem ordering online food, the chatbot takes less time navigating through a complaint registration, and the query is answered much faster.
▪ Quadrant II: High Functional Capability and Low Emotional Capability—This quadrant consists of digital assistants that are voice-based agents such as Alexa, Siri, and Cortana. Besides controlling home appliances such as speakers, televisions, lights, microwaves, and washing machines, they are now evolving with more functions such as driving and cooking assistance. The high functional capability enables digital assistants to project them as practical marketing tools according to Amazon’s deeper relationship strategy.
However, to be capable of performing several digital tasks is not enough to establish a friendly bond with the users. Smart speakers enabled with digital assistants offer a surface level of companionship based on the user’s desire for social contact with the speaker. Lack of self-disclosure and a weak empathetic design in digital assistants might be a gap.
The quadrant shows that digital assistants are the solution if a marketing activity requires high functional capability but can compromise on empathy. For example, ordering food from a voice assistant is more exciting and convenient than ordering from a food delivery website. 37 Therefore, digital assistants are highly useful when playing roles as shopping assistants in the decision stage of the customer journey because instant ordering and payments through voice commands can hasten the booking and transaction processes for the customers. Exemplifying this, another participant (male, 25 years old, with two and a half years of work experience) stated “I have used Alexa of Amazon as a companion. I have used it to ease my search for products while shopping on the online Amazon portal. Also, it helps build a conversation when needed, which helps [you] to feel good.”
▪ Quadrant III: Low Functional Capability and High Emotional Capability—This quadrant consists of ACs designed to engage users in empathetic conversations. ACs like Microsoft’s Xiaoice can recognize human emotions to engage them in lengthy discussions. AC’s accepting, understanding, and non-judgmental behavior allows users to disclose thoughts and feelings without hesitation.
Compared with digital assistants, empathetic chatbots are scant in functional capability. Consider Replika, which can engage users in romantic conversations but cannot turn on or off home appliances for the users. However, empathetic chatbots are the solution for marketing activities that require high empathy but low functional capability—for example, chatbots deployed as mental therapists. 38 Likewise, empathetic chatbots can be deployed as counselors and psychotherapists. Digital humans are futuristic versions of empathetic chatbots that can offer human-like counseling to users. Therefore, empathetic chatbots can also be deployed as sales advisors in the evaluation stage of the customer journey, where the user needs advice for the product or service or the price negotiation.
▪ Quadrant IV: High Functional Capability and High Emotional Capability—This quadrant consists of empathetic assistants designed to set up an elevated level of user companionship. Such ACs are high in functional and emotional capability and can offer users an experience of living with their favorite character, for example, Gatebox. 39 Empathetic assistants can take relationship marketing to the next stage. Brands can deploy representative ACs to serve customers, manifested as animated holographic agents portraying the brand character that can entertain customers as a therapeutic resource.
Firms can also give life to existing brand characters using artificial empathy with functional capabilities. For example, imagine McDonald Restaurant’s Ronald McDonald, Kellogg’s Tony the Tiger, Michelin Tire’s Michelin Man, PepsiCo’s Fido dido, and Coca-Cola’s Qoo come to life with AI’s help and serve as companions to the customers. This will enhance the customer-brand relationship, leading to long-term marketing goals such as customer retention, loyalty, and advocacy.
Therefore, empathetic assistants can be companions in the motivation and information search (pre-purchase) stage of the customer journey, enabling marketers to segment customers and promote products based on their needs. Once the brand character becomes the customer’s friend, it can navigate customer needs and wants by observing user emotions. Thus, organizations can sense the needs of prospective customers.
The classification matrix opens the scope for the role of ACs in the customer shopping journey, which has several strategic implications for the managers, as summarized in Table 3. The table explains When ACs can help users in the customer journey.
Strategic Implications of AI Companions in the Customer Journey.
Note: AI = artificial intelligence; AC = artificial intelligence companion.
The Future of Relationship Marketing
The following explains why ACs are significant to relationship marketing practices. 40 The theory of consumer engagement states that satisfying and emotionally connecting relations make partners engage with each other. 41 Successful entrepreneurs prioritize building an “emotional connection” with customers over simply chasing sales, ultimately fostering long-lasting and profitable loyalty. Employing ACs for customer engagement is a prudent and effective strategy to sustain a competitive advantage and enhance profitability. 42 Moreover, connection with multifaceted ACs like Alexa fosters long-term consumer-brand relationships and engages customers for extended periods. 43 ACs are the one-stop solutions for marketers to create, communicate, and deliver value to consumers. Existing marketing techniques collect consumer data from various channels and analyze it to predict consumer behavior. At the same time, deploying ACs allows brands to manage individuals’ preferences, convenience, likes, and dislikes more precisely.
Current marketing strategies rely on multiple channels to reach consumers, including websites, social media, television, and OTT (over-the-top) platforms like Netflix. Multiple channels can be effective in reaching a large number of consumers, but they might be expensive and time consuming to manage. Personalization for each consumer through multiple channels effectively becomes critical. ACs, on the contrary, can create a more personal and engaging relationship with consumers. ACs can engage with users more naturally and conversationally, building trust and rapport with users. For example, using website recommendation agents (RAs) becomes censorious for users. At times, the involvement of sponsorship practices enables RAs to propose biased suggestions toward sponsored products and deviates users from their preferences. 44 Whereas engaging users in free-flow conversations with ACs aligns users’ needs and preferences with the suggestions provided by ACs. Therefore, promotions and recommendations made by ACs are highly personalized, creative, and projected in real-time requirements.
An increasing number of ACs across industries around the globe indicates that our civilization is heading toward accepting AI for physical, mental, and emotional well-being. 45 In such a scenario, there is an urgent need for practitioners to think about building new relationship strategies using ACs. Empathetic ACs offer an irremediable urge to use AI to drive the consumer buying cycle and win the competitive advantage. An AC acts as a therapeutic resource for the users, whereas for marketers, it establishes a deeper relationship with the consumers. Figure 5 shows the implications of ACs at both ends. At the consumer’s end, implications include comfort, convenience, personalization, digital assistance, love, and care offered by the companion. At the organization’s end, it provides customer satisfaction, customer retention, word of mouth, customer advocacy, and customer loyalty.
Implications of “AI companion” for customers and organization.
Implications for Practice
As noted, task-specific AI agents on the firm’s website might be perceived as a biased recommender model by the customers as they are very limited in understanding customer problems, communicating, and recommending products or services. Therefore, free-form conversational ACs open vast opportunities for marketers to enhance relationship marketing strategies. The CDQ of ACs will position them as a practical tool for supporting users in the customer journey. There are many questions about the AC’s role. What capabilities enable an AC to establish user companionship? How to solve the problem of decreasing the interaction frequency of users with ACs? How should managers leverage each capability to utilize ACs in promoting businesses? Our study investigated these questions in depth to provide the following strategic guidelines to practitioners in relationship marketing with AI.
Understanding the Role of ACs in Relationship Marketing
By assessing existing models and comparing their compatibility with emerging technologies, practitioners can discover progressive approaches. 46 Firms need to understand the impact of ACs in the age of affective computing and generative AI to transform marketing. ACs, such as Alexa, Siri, Cortana, Replika, Mitsuku, Xiaoice, and ChatGPT, are embedded in the social network of consumers and influence their buying decisions. Once ACs become friends with the users, it opens a wide variety of opportunities for firms to understand, segment, and target the right consumers and position suitable products or services. For example, daily conversations between consumers and AI can generate large datasets that reveal user personalities, choices, needs, wants, and likings.
Elevated levels of companionship between AI and users can do wonders for marketing practices. The concept of selling involves identifying customer problems and needs to offer a solution through the firm’s product or service. ACs are often non-judgmental, allowing users to disclose personal information, such as challenges, difficulties, and problems. Providing solutions to customers through such ACs can benefit the consumers in two ways. First, this relieves stress and anxiety occurring because of the problem; and second, this allows them to see the right products and services that can make life easier. Businesses’ sales activity would increase and generate a larger customer base in a deeper relationship. Imagine the future models of digital humans, for example, UneeQ’s Sophie becoming more advanced and designed with a combination of generative AI, affective computing, digital assistance, object control, and transactional features. In such a changing scenario, managers must reshape marketing strategies considering ACs, as they can grab customers’ attention more than the earlier chatbots.
Following Capability-Focused Strategy to Utilize ACs in the Customer Journey
Intelligence is only one aspect of technology; however, consumer perception ties in with the remarkable capabilities of AI applications. Therefore, businesses should follow a capability-focused strategy. For example, firms tie up with the available AC providers or launch brand ACs to utilize capabilities in the customer shopping journey. ACs possess three foundational capabilities, such as conversational, functional, and emotional. Natural language processing, generative AI, context awareness, consistency, and coherence contribute to designing ACs with improved conversational capabilities. Firms can leverage this potential of ACs to engage customers with the technology for prolonged periods. For example, ChatGPT assists users in writing essays, poems, books, stories, news, letters, coding, and learning. This would make ACs more genuine, reliable, talkative, and informative. Firms can also benefit from such potential by encouraging consumers to use ACs in daily schedules, which would unfold the scope for deeper relationships between brands and consumers. The reimagined customer engagement in the AI age considers user-friendly interactions a strong pillar. 47
Voice recognition, voice command system, remote sensing, digital assistance, and task performance skills contribute to the design of ACs with functional capability. Firms can utilize the potential of ACs by allowing consumers to use them as shopping assistants that could showcase consumers with various products or services they search for and provide convenience to pay through voice commands for speedy transactions. The functional service experience offered by ACs enhances consumer engagement and more robust relationships. 48
Affective computing, feeling AI, perspective taking, empathetic concern, and emotional contagion contribute to designing ACs with emotional capability. Firms can leverage this potential of ACs to strengthen consumer-brand relations by offering an opportunity to users to live with their favorite brand characters and enjoy social or emotional conversations. Also, the emotional capability design of ACs can help users support them in the evaluation stage of the customer journey, as they would recommend products to the customers suitable to the needs and problems they are facing in real time. The three fundamental capabilities would lead to the perceived CDQ of ACs.
Overcoming the Decline in User Interaction Frequency with ACs
Extant studies found that users’ interaction frequency with ACs decreases over time. Engaging customers for longer periods is crucial to achieving profitable loyalty. We examined the problem through the lens of Mori’s uncanny valley theory and extended the graph to show how a victim of social isolation loses the emotional connection with the AC in the middle of relationship development due to a fall into an uncanny valley and, as a result, becomes lonely and socially isolated once again. The social alienation paradox exists because of single capability designs in ACs. For example, suppose Alexa can assist users with digital tasks. In that case, it does not qualify to be perceived as a better companion because it only offers users functional support but not emotional support. Similarly, empathetic chatbots such as Replica and Xiaoice can only provide emotional support to the users, but not a functional one.
In contrast, the paradox resolves when functional and emotional capabilities are combined in an AC, like in Gatebox. The user expects holistic ACs with a hybrid of conversational, functional, and emotional capabilities. Purely functional or emotional capabilities lead to an uncanny valley, wherein interest in the companion drops with time. The functional capabilities can pull users out of the uncanny valley, the emotional capabilities strengthen the bond between users and ACs, while the conversational capabilities enhance consumer engagement for prolonged interactions with memory-based conversations leading to profitable loyalty.
Our study suggests that managers design holistic ACs with a hybrid of multiple capabilities to achieve elevated levels of user-AI companionship and subsequently deeper relationships with the consumers.
Conclusion
Empathetic designs of ACs understand consumers’ emotional needs with the help of affective computing. With such empathetic skills, AI establishes companionship with users across industries and contexts, leading to proliferation across disciplines. However, extant literature focuses on intelligence types and lacks an understanding of the implications of such novel technology in marketing due to the associated challenges. ACs offer various benefits, but perceived risks can undermine the long-term vision for marketing. Intelligence is an inherent component of the technology, whereas customer perception is directly tied to AI’s capabilities. Therefore, we have examined the three fundamental capabilities leading to the perceived CDQ of ACs: conversational capability, functional capability, and emotional capability. Our study has showcased the merits and costs associated with establishing companionship with AI and highlights the challenge of reducing the frequency of user interactions with ACs over time. We have extended the uncanny valley theory by introducing the social alienation paradox. The paradox is a challenge for marketers to use ACs in business models. We recommend designing holistic ACs with a hybrid of multiple capabilities to prevent users from falling into the uncanny valley. The interplay between artificial companions’ capabilities in each stage of the customer shopping journey enhances marketing performance and builds deeper relationships with customers.
Footnotes
Notes
Author Biographies
Rijul Chaturvedi is a Senior PhD Scholar of Marketing at the Indian Institute of Management (IIM) Mumbai (email:
Sanjeev Verma is a Professor of Marketing at the Indian Institute of Management (IIM) Mumbai, India (email:
Vartika Srivastava is an Assistant Professor of Marketing at the Indian Institute of Management (IIM) Mumbai (email: