Information Systems Frontiers

https://doi.org/10.1007/s10796-021-10191-z

The Application of the Principles of Responsible AI on Social Media

Marketing for Digital Health
Rui Liu 1 & Suraksha Gupta 1 & Parth Patel 2

Accepted: 22 August 2021

# The Author(s) 2021

Abstract
Social media enables medical professionals and authorities to share, disseminate, monitor, and manage health-related information
digitally through online communities such as Twitter and Facebook. Simultaneously, artificial intelligence (AI) powered social
media offers digital capabilities for organizations to select, screen, detect and predict problems with possible solutions through
digital health data. Both the patients and healthcare professionals have benefited from such improvements. However, arising
ethical concerns related to the use of AI raised by stakeholders need scrutiny which could help organizations obtain trust,
minimize privacy invasion, and eventually facilitate the responsible success of AI-enabled social media operations. This paper
examines the impact of responsible AI on businesses using insights from analysis of 25 in-depth interviews of health care
professionals. The exploratory analysis conducted revealed that abiding by the responsible AI principles can allow healthcare
businesses to better take advantage of the improved effectiveness of their social media marketing initiatives with their users.
The analysis is further used to offer research propositions and conclusions, and the contributions and limitations of the study have
been discussed.

Keywords Responsible AI . Social media marketing . Digital health . Information sharing theory . Consumer trust theory .
Technology acceptance model

1 Introduction Twitter have all been identified as resources for patients to

communicate with each other. For example, Twitter users
The Internet has emerged as an imperative tool for people can mark their tweets as public to allow other users to read
seeking health knowledge from physicians, public health and retweet to others. According to two web-based patient
workers, and laypeople (Brownstein et al., 2009). Users of registries, 83 % of patients used social media to find out
social media particularly spend much of their daily time on information about their diagnosis or test results, 73 % of them
online applications where people discuss their wellbeing, read posts from rare disease groups or organizations, 67 % of
broadcast thoughts, experience and feelings about their health, them engaged in conversations about their diagnosis and 58
and generate many relevant data. Blogs, Facebook groups and % of them used social media to connect with others to seek
help (Rocha et al., 2018). Social media platforms offer public
health authorities and researchers the chance to collect
* Rui Liu health-related information generated by patients. In emergen-
[email protected] cy and catastrophe management, social media has also be-
come a significant communication route (Mirbabaie et al.,
Suraksha Gupta 2020). Along with AI and machine learning, social media
[email protected]
has been used in the prediction, detection, and treatment so-
Parth Patel lutions for mental health care via web and smartphone appli-
[email protected]
cations (D’Alfonso, 2020). Accountable technology, there-
1 fore, is consistently changing and co-shape human actions as
Newcastle University Business School, Newcastle University, 5
Barrack Road, Newcastle upon Tyne NE14SE, Tyne and Wear, UK it offers answers to the ethical questions of how people can act
2 (Peters et al., 2020).
Discipline of Management & Human Resources, Australian Institute
of Business, 1 King William Street, Adelaide 5000, South Australia, Researchers in the computing and information systems
Australia fields have done several pioneer practical explorations on
 Inf Syst Front

harnessing AI and machine learning to make digital inven- liability for actions of AI. Ghallab (2019) identified three im-
tions in improving user experience and optimize personalized portant risks in AI deployment (i.e., the safety of AI applica-
mental health care, as well as a therapeutic intervention. For tions, the security, and privacy for individual users, and the
example, to relieve the cost of long waits for clinic visits, social risks). Consequently, the current research and appliance
using AI instead of costly clinical staff, and to keep the “wor- in the field were restricted to a few separate elements of re-
ried well” from heading to the hospital, an artificially intelli- sponsible AI principles without a comprehensive understand-
gent chatbot on a mobile app was created to give British peo- ing. Secondly, despite the widespread use of social media in
ple diagnostic advice on common ailments without human promoting, the application of responsible AI in facilitating
interaction (Olson, 2018). It was also a good replacement for digital health through social media is scarce; thirdly, associat-
the health advice telephone line which couldn’t show how ed practical investigations of responsible AI are absent and
many patients take the advice, and AI bots could track that therefore the research lacks an empirical foundation. This pa-
data. Natural language processing (NLP) is one of the per has endeavored to fill in the gaps by proposing a set of
AI-powered tools in analyzing the language of humans and responsible AI principles and putting eight principles into dig-
solutions to mental health. Machine learning and NLP tech- ital health practices that are critical in determining the effec-
nologies have explored the potentials for extracting useful tiveness of responsible AI.
health information from substantial data of the Internet and
have earned substantial achievements (Dredze, 2012). 1.2 Theoretical Issues and Limitations
User-generated content comes from a personal digital device
and social media interactions, and the individual’s ‘digital While many cross-disciplinary experts have acknowledged
exhaust’ is continuously creating a growing reservoir of data the concerns of blameworthy irresponsible AI at a fast rate
for NLP studies and can be mined for behavioural and mental and published many AI principle models, the attempts of these
health insights, deep and real-time data analysis potential cul- principles to translate into AI practices is still in their infancy
tivation (D’Alfonso, 2020). The widespread use of social me- (Benjamins, 2020; Cheng et al., 2021; Scantamburlo et al.,
dia combined with AI, help people with behaviour and mental 2020). The reason at its core relies on their abstract nature that
illness to be effectively treated at a relatively low cost by makes practitioners feel difficult to operationalize
filling in the gap between individuals’ demands for healthcare (Scantamburlo et al., 2020), and companies might overlook
resources and those with abundant scientific and efficient ac- the financial rewards of responsible AI and treat it as the sole
cess. However, social media users may perceive risks and way to avoid risks (Cheng et al., 2021). Also, there is no
uncertainties when sharing their health data on social media consensus about which principles should be included for re-
to achieve better health. The level of user technology accep- sponsible AI and how to practically deal with these issues in
tance is determined by some factors including perceived use- organizations (Gupta et al., 2020; Benjamins, 2020).
fulness, perceived ease of use, behavioural intention, and oth- The majority of the attention has mainly focused on the
er contextual factors. unintended negative impacts of responsible AI on ethical is-
sues rather than intended and controllable impacts because the
1.1 Research Gaps intended motivation of developers seems less likely to release
a new technology blindly (Peters et al., 2020; Wearn et al.,
In terms of research gaps, this research is important for three 2019). Relatively little attention has been paid to understand
reasons. Firstly, even though how to manage AI responsibly proactive organizational ethical efforts. Without a detailed un-
has invoked some debates in broad scope, a systematic dis- derstanding of specific AI tools from a third party, an organi-
cussion about responsible AI principles is limited. In a study zation may trigger biases that negatively impact the brand and
for examining ethical AI principles in maintaining their de- creates compliance issues. For example, Eitel-Porter (2021)
ployment in organizations, Eitel-Porter (2021) focuses solely clarified three categories of responsible AI: compliance and
on the trust, fairness, and privacy ingredients of AI, and does governance, brand damage and third-party transparency.
not include data security. Arrieta et al. (2020) established that Specifically, biases in training data may lead to recruitment
fairness, accountability, and privacy should be regarded when apps to favour one gender which breaches anti-discrimination
bringing AI models into practice. Under increasing use of AI laws; AI algorithms may prioritize the delivery services in
in high-stakes decision-making, another research on the de- affluent areas, which breaches social norms and taboos.
sign of a contestation process has found its effects on the
perception of fairness and satisfaction; it also found that a lack 1.3 Managerial Research Issues
of transparency can be due to design challenges. (Lyons et al.,
2021) Lima & Cha (2020) discussed three notions of respon- Despite the opportunities and benefits of AI, experts re-
sible AI, including blameworthiness, accountability, and vealed various risks regarding the interests of the
Inf Syst Front

organizations and their stakeholders (Clarke, 2019). The 1.5 Theoretical and Practical Contributions
concept of responsible AI has been stated which seeks to
support the design, implementation, and use of ethical, Thus, this research constructs and proposes a conceptual mod-
transparent, and accountable AI solutions, aiming at re- el that attempts to elaborate on the effective responsible AI
ducing biases, facilitating fairness, equality, interpretabil- activities from both the organizational level and consumer
ity, and explainability (Trocin et al., 2021). The hope for level, by surveying the responses of healthcare managers
the development of a Good AI Society has been proposed and employees towards their expectations, feelings, experi-
accordingly (Floridi et al., 2018). They acknowledged the ences, and concerns to validate the model. The authors of this
impact of AI technology for promoting human dignity and research have incorporated three typical frameworks of re-
human flourishing, and also identified sequent problems sponsible AI principles practically from a company
of privacy issues, unfairness, unreliability, and unsafety, (Microsoft AI, 2020) and theoretically from two articles
maleficence which emanated as detrimental barriers for (Clarke, 2019; Floridi et al., 2018) into a discussion, to exam-
the sustainability of organizations. For instance, mental ine whether the existing responsible efforts made by AI de-
health care has raised particular ethical and legal consid- signers, operators and other users in the healthcare industry
erations as well as the need for regulation by its nature, are liable or not. Further from the impacts of responsible AI
such as privacy v iolation and rooted prej udice activities on social media marketing efforts and influences of
(D’Alfonso, 2020). Ethics have an impact on people’s these efforts on digital health, this new developed conceptual
digital health in terms of privacy, fairness, inclusiveness, model would also help better understand the moderating role
transparency, accountability, privacy, security, reliability, of consumer trust and data quality in the influencing mecha-
and safety (ibid.). The expectation and the response of nism, because most of the existing research on consumer trust
users are changing. As a result, the design, assessment, is focusing on the trust of online information, while this paper
and use in diverse industries of ethical AI should be would switch the focus on whether consumer trust of the plat-
ongoing and iterative. Floridi et al. (2018) proposed an forms and technology might take effects on social media mar-
ethical framework for a Good AI Society. They construct- keting of digital health in organizational level. Additionally,
ed this framework from the researchers’ point of view, relevant research on the theory of information sharing use in
hoping the framework can be undertaken by policymakers inspecting social media performance is so scarce and requires
and each stakeholder. The managerial research issues in further consideration. To seek healthcare expertise views, in-
this research are to examine the solutions to ethical prob- dustrial professionals, policymakers, and others can take ac-
lems faced by AI and its use in social media. This re- tions and make changes, set up policies, regulations, and laws
search seeks to bridge the responsible AI principles and more responsibly in the public information sharing by stake-
theories to business practices particularly in the healthcare holders such as patients and clinicians. Consumers would feel
industry. more secure about sharing their health data on social media
since it is no longer risky when they leave a footprint on social
media, to achieve better health more cheaply and convenient-
1.4 Research Questions, Aims, and Objectives ly. This paper also showcases the opportunities and challenges
under the responsible AI paradigm.
In light of the abovementioned gaps, this paper attempts to
answer the following research questions:
RQ1) How organizational appliance of responsible AI 2 Literature Review
principles in practice can contribute to the digital health
maintenance of social media consumers? And 2.1 What is Responsible AI?
RQ2) How does consumer trust and data accuracy mod-
erate the relationship between responsible AI principles Responsible AI has been used interchangeably with ethical AI
and activities? or the responsible use of AI. Eitel-Porter (2021) defined the
term as the practice of using AI with good intention, to em-
To be specific, the aims of objectives can be clarified as power employees and businesses and create a fair
follows: environment for customers and society, ultimately enabling
(1) To construct a set of responsible AI principles which can organizations to generate trust and bring AI deployments to
guide organizational responsible AI activities. scale. Taylor et al. (2018) regarded responsible AI as an um-
(2) To address the ethical dilemma of responsible AI solu- brella term, which investigates legal, ethical, and moral view-
tions in social media. points of autonomous algorithms or applications of AI that
(3) To provide reliable evidence to guide professionals in may be crucial to safety or may impact people’s lives in a
improving the delivery of digital health. disruptive way. AI incorporates a broad range of techniques
 Inf Syst Front

and approaches of computer science to simulate human intel- AI. Contractor et al. (2020) have built a framework and of-
ligence in machines that are programmed for thinking like fered suggestions to relieve the concerns about inappropriate
humans and mimic human behaviours, capable of performing or irresponsible use of AI through licensing on software and
tasks. During the past 80 years after World War II, the ad- data to legislate AI usage. While Eitel-Porter (2021) asserted
vances of AI stimulate the invention, innovation, and invest- that responsible AI needs mandated governance controls in-
ment in computing skills (Clarke, 2019). Some AI technolo- volving methods for managing processes and enforcing audit
gies are relatively longstanding, such as online robots for trials. All of these articles solely concentrate on the abstract
counselling, while others are more recent for example compu- conceptual level of discussions such as guidance and
tational modelling used by social media companies to identify principles.
users at risk of self-harm. The identification of AI technology In the AI-enabled social media marketing field, an overload
and associated appliances in different situations with the of knowledge, false reports, lack of signal specificity, and
changes of markets and society engendered both opportunities sensitivity to external forces such as the interest of the media
and threats to organizations. might greatly limit the realization of the capacity of health care
AI technologies may be underused, overused or misused, for public health practices and clinical decision-making
associated with fear, ignorance, misplaced concerns, or exces- (Brownstein et al., 2009). The researchers also point out the
sive reaction (Floridi et al., 2018; Meske et al., 2020). user privacy issues (Dredze, 2012; Kapoor et al., 2018). Users
According to Eitel-Porter (2021), unintended negative conse- of social media hold privacy and ethical issue concerns and
quences for organizations may occur when AI applications expectations towards AI, such as algorithms that infer unstat-
have been done without care. For example, faulty or biased ed user demographics or diagnoses from public data (Dredze,
AI applications which breach the risk compliance and gover- 2012). Responsible AI should tackle the tension between in-
nance may damage the brand. The use of AI, therefore, should corporating the benefits and mitigating the potential harms of
be coupled with good innovation and positive execution of AI and avoiding the misuse and underuse of AI (Floridi et al.,
this technology abiding by moral principles and social norms. 2018). Integrating ethics into AI allows organizations to take
The larger number of people who enjoy the opportunities and advantage of the social norms, values, and culture, participate
benefits of AI, the more responsibilities are essential in terms in those socially acceptable or preferable, legally unquestion-
of what type of AI to develop, how to use it, and who can use able activities, and prevent and minimize costly mistakes
it properly. There is a great need to guide how AI should be (ibid.). Hence, examining the responsible use of AI in social
permitted, how it can be used responsibly, and when it should media emerges as an essential activity for AI-associated orga-
be discouraged or forbidden raises without precedents nizations and researchers.
(Gooding, 2019). Many common pitfalls raise risks for an Responsible AI is about answering who determines which
organization: rushed development, a lack of technical under- are the alternatives and how to implement for ethical
standing, improper quality assurance, use of AI outside the decision-making by AI systems, consistent with societal, eth-
original context, improper blends of data, and reluctance by ical, and legal requirements. Existing approaches to the imple-
employees to raise concerns (Eitel-Porter, 2021). mentation of ethical reasoning can be divided into three main
Consequently, proper use and the proposal of accountable categories: top-down approaches, bottom-up approaches, and
guidelines of AI have emerged as crucial concerns for the hybrid approaches. Top-down approaches are to implement a
public and the organizations. given ethical theory and apply it to a particular case;
There are a number of factors that impact the behaviour of bottom-up approaches are to aggregate sufficient observations
AI systems, so the term ‘responsible AI’ has been used in of similar situations into a decision and infer general rule from
different fields, aiming at mitigating bias in data collection these independent cases; hybrid approaches combine the ben-
and design of the algorithm and engendering better interpret- efits of bottom-up and top-down approaches in support of a
ability and explainability (Contractor et al., 2020). careful moral reflection (Singer, 2020). From a deontological
Nevertheless, although the significance of responsible uses view, this paper leverages the top-down approaches by apply-
of AI has been realized and discussed in the past few years ing the current responsible AI frameworks to the social media
and concerning diverse responsible AI principles and health cases and then describes what people should do in this
frameworks, the research on responsible AI remains at the specific situation, judge the ethical aspects or ‘goodness’ of
theoretical stage, and it seems to be difficult to apply it to behaviours in AI systems and to ensure ethical acceptance and
the practical hierarchy or to apply it to industries broadly. social acceptance for the ethical reasoning by AI systems.
Another problem with responsible AI research is that most
responsible AI guidelines are useful and can help shape 2.2 The AI Ethical Principles
policy but are not easily enforceable. Arrieta et al. (2020)
presented a theoretically oriented analysis for concepts, tax- There are many researchers and business organizations that
onomies, opportunities, and challenges towards responsible have produced different categories of responsible AI
Inf Syst Front

principles, approaches, and practices throughout the existing prioritize the autonomy of people throughout the design and
AI practice and literature (Arrieta et al. 2020; Benjamins et al. execution of AI. Table 1 comprehensively depicted and sum-
2019; Clarke, 2019; Ghallab, 2019; Lima & Cha, 2020; Lyons marizes the three important responsible AI principles model
et al. 2021). For example, Arrieta et al. (2020) stated that and organize similar concepts together to help build a better
fairness, ethics, transparency, security and safety, accountabil- understanding of the existing frameworks. This table has re-
ity and privacy should be considered as AI models that need to moved some original principles from the three groups of re-
be implemented in the real world and this would bring a grad- searchers because they were not relevant to this research and
ual process in increasing corporate awareness around AI prin- also because there might be overlapping areas with other prin-
ciples. According to the European Commission’s ‘Ethics ciples, and so the table has only retained eight essential prin-
Guidelines for Trustworthy AI,’ trustworthy AI should be ciples which were fairness, inclusiveness, reliability and safe-
lawful, ethical, and robust both from a technical perspective ty, transparency, privacy and security, beneficence,
and social environment (European Group on Ethics in Science non-maleficence, and autonomy.
and New Technologies, 2018). Beijing Principle, which is the
first principle of artificial intelligence for children, has been 2.3 Social Media as Channels for Disseminating Digital
stated in 2020 and has proposed four values including dignity, Health Information
growth, fairness, and children first. The Beijing principle sug-
gests that the development of AI should protect and promote Three directions of AI use in social media are stated as fol-
the physical and mental safety of children, protect them from lows, involving discovering health topics, bio-surveillance
physical and mental violence, injury or abuse, neglect or neg- monitoring and discovering self-management information.
ligent treatment, maltreatment, or exploitation. And responsi-
ble AI should help combat child trafficking, indecency, and 2.3.1 Discovering Health Topics
other crimes. The collection of information on children should
ensure their guardians’ informed consent and avoid illegal In disease surveillance and epidemic detection, control of
collection and abuse of children’s information. Responsible chronic diseases, behaviour monitoring, and public health
AI systems should ensure that children and their legal guard- communication and education, health intelligence contributes
ians, or other caregivers have the rights to consent, refuse, to these areas (Shaban-Nejad et al., 2018). Public-health offi-
erase data, and revoke authorizations (BAAI, 2020). cials can leverage social media websites that produce real-time
Clarke (2019) proposed ten principles and claimed that data about a daily population of advanced events to discover
these principles can be used as a basis for particular AI arte- health issues, including expected seasonal events, such as in-
facts, systems, and applications in diverse AI industries. fluenza, allergies, disease outbreaks, food poisoning, or a bio-
Microsoft has launched the Office of Responsible AI (ORA) chemical contaminant (Dredze, 2012). For example, one re-
and the AI, Ethics, and Effects in Engineering and Research search uses supervised machine learning to discover the types
(Aether) Committee to put responsible AI principles into prac- of health topics discussed on Twitter, and how tweets can
tice (Microsoft AI, 2020). Microsoft has claimed six respon- augment existing public-health capabilities. And by examin-
sible AI principles, including fairness, inclusiveness, reliabil- ing the disease words, symptoms, and treatments with the
ity and safety, transparency, privacy and security, accountabil- ailment, supervised machine learning uncovers 15 ailments
ity (ibid.). Another twelve researchers have created five major including headaches, influenza, insomnia, obesity, dental
principles of responsible AI by incorporating four core prin- problems, and seasonal allergies, such as the H1N1 virus
ciples commonly used in bioethics (i.e., beneficence, (ibid.). Besides, social media content could be a valuable
non-maleficence, autonomy, and justice) with a new principle source of information for predicting or comparing different
named explicability that combines intelligibility and account- data sets during different periods (Stieglitz et al., 2020).
ability (Floridi et al., 2018). They believe that the four bioeth-
ical principles can adapt well to the ethical challenges of AI. 2.3.2 Bio-surveillance Monitoring
The key contribution of this statement of responsible AI prin-
ciple is that they involve the original intention of the creation The increasing usage and public participation of social media
of any technology, which should be a benefit for and no harm platforms have provided the government with a chance to get
to mankind. Particularly for digital health, organizations access to public opinions through social media monitoring and
should respect the principles of bioethics. Although people controlling of governmental policies for the welfare of the
are passive receivers in the digital world which means they general public (Singh et al., 2020). Monitoring online health
do not have the power to refuse embedded technologies, in- discussion offers valuable insights into public health condi-
formation, and interruption, organizations still need to tions that are not valuable in terms of more traditional methods
 Inf Syst Front

Table 1 A summary of three important responsible AI principles model

Elements Clarke (2019) Microsoft AI (2020) Floridi et al. (2018)

Justice/fairness Process and procedural fairness and AI systems should treat all people fairly and Using AI to correct past mistakes such as
transparency should be fulfilled. produce fairness rather than reinforce bias eliminating unfair discrimination and to
and stereotype to society create shared or sharable benefits without
creating new harm.
Inclusiveness - AI systems should intentionally engage -
communities.
Reliability and Embedded quality assurance. AI system should be consistent with -
safety designers’ ideas, organizational values, and
principles. It applies to any products of the
company.
Transparency Ensure accountability (i.e., each AI systems should be understandable, and The relationship between humans and this
entity is discoverable) for legal people can understand behaviors of AI, transformative technology should be readily
and moral obligations designers open to users with why and how understandable to the average person.
they create the system.
Privacy and - AI systems should be secure and respect -
security privacy through considering data origin and
lineage, data use internal and external.
Beneficence Consistency with human values and - The original motivation of creating AI
human rights. technology is to promote the benefits or
well-being of humans and the planet with
dignity and sustainability.
Non-maleficence Safeguards for human stakeholders - Be cautious against the potentially negative
at risk should be provided and consequences of overusing or misusing AI
replace that inhumane machine technologies, for example, the prevention of
decision-making. infringement on personal privacy, even
worse as the AI arms race. Accidental or
deliberate harm should be taken seriously
whether from the intent of humans or the
unpredicted behavior of machines.
Autonomy Human ceding power to machines; - A principle that the autonomy of humans to
but all stakeholders have legal make decisions should be protected rather
and moral obligations to assess than delegating too much to machines.
the impacts of AI

Sources: Clarke (2019). Principles for Responsible AI. https://arxiv.org/abs/2101.02032. Accessed 1 November 2020
Floridi et al. (2018). AI4People-An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations. Minds and
Machines, 28(4), 689–707
Microsoft AI, 2020. Responsible AI. https://www.microsoft.com/en-us/ai/responsible-ai?activetab=pivot1:primaryr6. Accessed 4 October 2020

which depend on manual transcripts from clinicians and other 2.3.3 Sharing Self-management Information
healthcare workers (Brownstein et al., 2009). Global Health
Monitor is one of the web-based systems for detecting and Social media data forms new public-health capabilities,
mapping infectious disease outbreaks (Doan et al., 2019). particularly for those people who are reluctant to discuss their
The system can analyze and classify English news stories issues with healthcare workers (Dredze, 2012). For example,
from news feed providers for topical relevance and use negative perceptions and discrimination towards persons with
geo-coding information to help public health workers to mon- mental illness are substantial and widespread (McClellan et al.,
itor the spread of diseases in a geo-temporal context. Besides, 2017) which hinders those persons to seek help, however, so-
social media platforms provide users’ location information for cial media information provides open resources to them.
researchers to execute bio-surveillance, for instance, to Among all of the health issues, mental health disorders are
direct vaccine suppliers to the areas and populations affecting a substantial portion of people worldwide. More than
(i.e., demographic groups) where they were most need- 80 % of people globally are experiencing mental health condi-
ed. Overall, social media has already become one of the tions, including individuals who are experiencing neurological
most usefu l sou rces for g ove rnmen ts to m ak e and substance use disorders (Ghebreyesus, 2019). Social media
bio-surveillance monitoring while AI holds great prom- platforms have become a hidden place for people with mental
ise for improving the delivery of health services in health disorders to share and absorb associated mental health
resource-poor areas (Wahl et al., 2018). experiences, feelings, and medical suggestions.
Inf Syst Front

2.4 The Appliance of Social Media and AI for Health indicate our psychological states, therefore, the NLP tech-
Care nique enabled by AI and machine learning technologies has
been applied to examine the associations between language/
Traditionally, public health requires collecting clinical data voice feathers and mental health (D’Alfonso, 2020). By ana-
aggregation from clinical encounters, which are expensive, lyzing linguistic characteristics of social media contents, re-
time-consuming, and slow (Dredze, 2012). However, the ad- searchers can produce a machine learning model that can be
vances of AI capabilities in computational and data sciences used to forecast an individual’s mental health earlier than tra-
assist us to extract and analyze social media data to estimate ditional methods, for example, another important issue in the
the incidence and prevalence of different health conditions, as Digital Therapeutic Alliance (DTA), which incorporates a
well as related risk factors (Shaban-Nejad et al., 2018). The therapist into a patient journey when offering therapeutic in-
application of AI to participatory health informatics can en- terventions through a smartphone, a web page, or a sophisti-
compass the use of physiological data used from text-related cated conversational agent (ibid.). Therefore, mental health
data, electronic health records, social media, wearable de- professionals can tailor their efforts accordingly.
vices, and clinical trials. Current literature has done piles of
empirical research on the appliance of social media and AI for 2.5 Consumer Trust Theory
improving public health (see Table 2).
Digital technologies are used in digital health initiatives People are highly motivated to scrutinize health information
including various types of technology, including those de- online, and the credibility of this health information judgment
signed for information sharing, communication, clinical deci- is correlated more with information characteristics rather than
sion support, ‘digital therapies’, patient and/or population personal health status (Ye, 2010). Consumers’ lack of trust
monitoring and control, bioinformatics and personalized med- during the online navigation process is manifested in many
icine, and service user health informatics (Coiera, 2015). A aspects, such as their concerns that the personal information
survey showed that the vast number of experts, more than 75 would be sold to third parties beyond their knowledge or per-
%, use Twitter data, and more than half prefer to use regres- mission (Hoffman et al., 1999). Although many organizations
sion algorithms to do social media prediction, but not all fore- are aware that lack of trust might lead to consumers’ unwill-
casting models can predict accurately, and prediction appears ingness to engage in the relationship exchange on social me-
to be reliable on the affiliated field (Rousidis et al., 2020). dia, they are still reluctant to ask consumers to opt-in. Because
Although the social media population comprises only a spe- of these worries, most consumers would opt out of informed
cific fraction of the population, the reach of its posts can cover consent. And, even though such websites have told users ex-
broader impacts through social multiplier effects (McClellan plicitly that they are tracked and recorded, in many cases,
et al., 2017). consumers are not allowed to reject some permissions if they
Several existing pieces of literature have researched need to use the social media websites.
health-related studies associated with AI use on social media The advancement of AI capacities enables the flourish of
through the following ways: simple frequency analysis, con- data mining and data warehousing opportunities. Social media
tent analysis, semantic analysis, supervised learning, and a gather an unprecedented number of personal data that raises
major analytic approach (i.e., time series analysis and forecast- concerns to those consumers with profound distress or
ing techniques) (Briand et al., 2018; McClellan et al., 2017; particularly, mental health crisis (Gooding, 2019). Personal
D’Alfonso, 2020; Rousidis et al., 2020). The time series anal- transaction information such as identity, credit history, ad-
ysis has some limitations such as the lack of ability to recog- dresses, and other information such as searching history, in-
nize sarcastic or humorous tweets, but it could be refined if ternet sited visited, preferences, and even illness information
combines with sentiment analysis (McClellan et al., 2017). are leveraged along with each click of consumers. Hence, for
Briand et al. (2018) combined supervised learning and infor- website designers, runners, and market practitioners, respect-
mation retrieval methods to analyze user-generated content in ing consumers’ rights to data ownership on the Internet can be
social media for the early detection of mental health condi- the priority in earning consumers’ trust. A comparison study
tions. McClellan et al. (2017) applied the autoregressive inte- among three European countries revealed that taking con-
grated moving average (ARIMA) model to identify deviations sumers’ perception of trust into account is important in devel-
of social media data from the predicted trend in real-time and oping, launching, and marketing health-enhancing,
forecast time series data. They developed a model to identify non-edible products (Puhakka et al., 2019). Admittedly, con-
periods of heightened interest in suicide and depression topics sumer behaviour theories have well acknowledged that con-
on Twitter. sumers perceive risks in their behavioural intention. Rather,
AI is being applied in mental health by natural language this essay will examine the role of consumer trust level in
processing (NLP) of clinical texts and social media content. influencing the effectiveness of responsible AI principles
The language people use, and our vocalizations are believed to complementation. If consumers do not trust responsible AI
 Inf Syst Front

Table 2 Select studies on combined social media and AI use on health issues

Study Sample Research context Objects of analysis Position of Research Major findings
responsible AI gaps/Theoretical
principles in the contributions
conceptual
model

Rocha et al. A total of 103 Rare disease Two online patient Privacy. There is a paucity of • There is broad
(2018) responses community and registries Autonomy. literature variability between
from genetic testing Reliability. characterizing the individuals’ privacy
GenomeConn- potential for preferences, according
ect and communication, to experiences,
Simons VIP networking, privacy concerns, and
registry and membership adaptation to their
participants preferences, and diagnosis or genetic
support needs for rest results.
people with rare • Patients wish to have
genetic diagnoses. some control over the
This preliminary visibility of the
work could inform the information they
design of more robust share.
and nuanced research • Genetic counselors
of rare disease should provide
communities patients with guidance
collectively or about reliable social
specific rare disease media resources for
communities information.
individually.
Denecke 22 articles and 12 Participatory Seven databases and Transparency. Although AI for • AI may require the
et al. clinical trials health online forum Beneficence. supporting design to be embedded
(2019) involving AI informatics (clinicaltrials.gov) Privacy. participatory health is deeply or even
in still in its infancy, invisibly in patients’
participatory there are a number of daily routine.
health contexts important research • The analysis of social
priorities that should media data with AI can
be considered for the provide new insights
advancement of a into patient health
field such as the beliefs and
psychosocial perspectives on their
wellbeing of health, healthcare use,
individuals and wider and efficacy and
acceptance of AI into adverse effects of
the healthcare drugs and treatments.
ecosystem. •The ethical and practical
privacy issues using
healthcare data (such
as medical images,
biological data,
experiential reporting,
and physiological
data) need to be
urgently addressed by
health systems,
regulators, and society.
McClellan 176 million Depression or Twitter activities: Beneficence Although ARIMA • Spikes in tweet volume
et al. tweets from suicide expected response to models have been following a behavioral
(2017) 2011 to 2014 planned behavioral used extensively in health event often last
with content health events and other fields, they have for less than 2 days.
related to unexpected response not been used widely • By monitoring social
depression or to unanticipated in public health. The media
suicide events findings indicate that communications and
the ARIMA model is timing dissemination
valid for identifying of information about
periods of heightened mental health,
activity on Twitter prevention and
Inf Syst Front

Table 2 (continued)

Study Sample Research context Objects of analysis Position of Research Major findings
responsible AI gaps/Theoretical
principles in the contributions
conceptual
model

related to behavioural treatment initiatives

health. can be taken by
government agencies
and public health
organizations.
Tutubalina, 217,485 reviews High blood WebMD.com, a health Reliability and Medical applications, • While neural networks
and from authors pressure, pain, information services safety demographic in this kind of
Nikolen- tagged as depression, website that provides information regarding NLP-related problems
ko, ‘patient’ chronic trouble credible information, the authors of reviews perform better in terms
(2018) sleeping, supportive such as age and of the
attention deficit communities, and gender is important, classification/-
disorder with in-depth material but existing studies regression objective,
hyperactivity about health subject usually either assume topic models learn and
that this information provide extra
is available or information that may
overlook the issue lead to interesting
entirely. The study observations relevant
found that to the underlying
convolutional neural healthcare application.
networks perform
best in predicting
demographic
information and topic
models provide
additional
information and
reflect gender-specific
and age-specific
symptom profiles.
Ahmed 214,784 tweets H1N1 pandemic Twitter data were Reliability and Novel insights were • Twitter data could be
et al. from 28 April retrieved from the safety derived on how users utilized by library
(2009) to 29 April Twitter Firehose API communicate about professionals for
2009 via a licensed reseller, disease outbreaks on developing a better
Visibrain (n.d.) social media understanding of
platforms. public views on
The study also provided health-related topics.
an innovative • Using an in-depth
methodological qualitative method
contribution. such as thematic
analysis when
analyzing social media
data may lead to
greater insights.
• Misunderstandings of
medical advice can
lead to dangerous
consequences and
must be understood
carefully.
Sumner 95,555 social Suicide Twitter, YouTube, Beneficence, Social media messages • Novel online risks to
et al. media posts Reddit, Tumblr, and Non-maleficence and online games mental health, such as
(2019) and articles blogs, forums, and promoting suicide are pro-suicide games or
about an news articles. a concern for parents messages, can spread
alleged suicide and clinicians. The rapidly and globally.
game were study provided a • Better understanding
collected better understanding social media and Web
of the degree to which data may allow for
 Inf Syst Front

Table 2 (continued)

Study Sample Research context Objects of analysis Position of Research Major findings
responsible AI gaps/Theoretical
principles in the contributions
conceptual
model

social media data can detection of such

provide earlier public threats earlier than is
health awareness. currently possible.
Booth et al. Monthly Mental health The inclusion of Twitter Non-maleficence The study provided • The 2012 Bell Let’s
(2018) outpatient awareness and into the 2012 Bell important Talk campaign was
mental health stigma Let’s Talk campaign. methodological temporally associated
visits reduction implications for with an increase in the
researchers wishing to rate of mental health
ascertain the efficacy visits among Ontarian
of social media and youth.
other digitally
enabled media
campaigns operated at
population level.
The current 25 social media Responses of Weibo, WeChat and Eight principles A systematic discussion
study practitioners interviewees other health about responsible AI
from health regarding communities, social principles is limited;
industry. responsible AI media pages of the application of
medical consultation responsible AI in
software. facilitating digital
health through social
media is scarce;
associated practical
investigations of
responsible AI are
absent and therefore
the research lacks an
empirical foundation.

practices, the efforts made by the managers would be in vain. capabilities influences the willingness as does life satisfaction
And this assumption is rooted in that even though responsible (Zaheer & Trkman, 2017). IST draws its roots from social
AI practices play a complementary role in determining users’ exchange theory (SET) issues (e.g., trust, commitment, reci-
actual behaviour, they may also have a deep impact on the procity, and power) and social psychological factors of the life
activities’ performance. satisfaction of individuals (i.e., attitudes, feelings, and
self-identity) to examine the effects on people’s intentions to
2.6 Theory of Information Sharing share information (Constant et al., 1994; Wu et al., 2014;
Zaheer & Trkman, 2017). And the IT infrastructure capability
The information-sharing theory (IST) is based at the outset assessable to relevant stakeholders facilitates their willingness
that “organizational culture and policies as well as personal to share quality information (e.g., timely, accurate, adequate,
factors that can influence people’s attitudes about information complete, and reliable) on such platforms (Zaheer & Trkman,
sharing” (Constant et al., 1994). The purpose of IST is to 2017). Task interdependence, perceived information useful-
understand the factors that support or constrain information ness, the user’s computer comfort, information ownership,
sharing in technologically advanced organizations (Jarvenpaa and the propensity to share are strongly related to the individ-
& Staples, 2000). The human attitude (i.e., willingness to ual’s use of collaborative media (Jarvenpaa & Staples, 2000).
share, its antecedents, and role) contributes to improving in- The more interdependent a person’s work is on others, the
formation sharing quality. According to a social- higher the needs of self-interest and reciprocity are and there-
psychological study about the willingness to share informa- fore people are more likely to share (ibid).
tion, trust, commitment and reciprocity were considered as There are three types of information sharing: operational,
important antecedents that influence the willingness to share tactical, and strategic (Rai et al., 2006). Operational informa-
information with varying effects; access to proper IT tion sharing concerns managing the flows of materials,
Inf Syst Front

components, and finished goods in a way to optimize how responsible AI principles impact initiatives of responsible
production-related activities; tactical information sharing fo- AI from the consumer level to the organizational level.
cuses on collaborative partners to improve decision quality;
strategic information sharing incorporates group members in a 2.8 Technology Acceptance Model
strategic form for gaining competitive value on the
industry-wide structure. The technology acceptance model (TAM) was introduced in
The theory of information sharing was used in social media 1986, revised in 1989, and has evolved as a significant model
analysis. Social media is primarily used as a personal learning in understanding and predicting human behaviour concerning
space but is also used as a knowledge management tool and to the potential acceptance or rejection of the technology (Davis,
develop communities for information sharing. Privacy has 1985; Davis, 1989; Lee et al., 2003; Rauniar et al., 2014;
been identified as the main concern for users of a personal Marangunić & Granić, 2015; King & He, 2006). This model
learning space from the disclosure of potential benefits, assumes that an individuals’ technology acceptance is deter-
long-term use, the variety of personal artefacts to a wide range mined by two major factors: perceived usefulness (PU) and
of audiences (Razavi & Iverson, 2006). For example, accord- perceived ease of use (PEOU) and a dependent factor behav-
ing to the survey, participants confirmed that cautious feelings ioural intention (BI) (King & He, 2006). The intention, in turn,
emerged when they join a new community and the trust level is determined by an individual’s attitude (A) towards the tech-
improves which leads them to share more freely after a longer nology and perceptions concerning its usefulness (Szajna,
period (ibid.). To create privacy management mechanisms for 1996). Some researchers have examined the TAM model in
personal learning, spaces should be based on users’ mental the health care realm. For example, Holden and Karsh (2010)
model of information privacy (i.e., privacy concerns, privacy reviewed the application of TAM by analyzing over 20 studies
strategies, and privacy needs). Additionally, the understand- of clinicians using health IT for patient care. They found that
ing and practice of information sharing have become increas- the relationship between PU and BI or actual use of health IT
ingly significant for organizations to keep competitive and is significant in each test, which implied that to improve the
facilitate profitability (Hatala & George, 2009). This paper use and acceptance, the health IT must be perceived as useful.
integrates the technology acceptance model, consumer trust Designers, buyers, and other stakeholders involved with AI
theory, and information sharing theory to answer the afore- are advised to use TAM to assist the design or purchasing
mentioned research questions in Section 1. process, training, implementation, and other activities
(Holden & Karsh, 2010). Another research on social media
user’s attitudes found that utilitarian orientations of PU and
2.7 Integrating Consumer Trust Theory and trustworthiness of a social media site are crucial to use inten-
Information Sharing Theory tion and actual use, and user engagement on social media
needed to be considered (Rauniar et al., 2014). Similarly, re-
There are many overlapping areas between consumer trust searchers found that encouraging a positive attitude of the
theory and information sharing theory. Trust was considered technology’s usefulness toward using technology is crucial,
as an important antecedent of information sharing theory that which indicated that information sessions and training on tele-
influences people’s willingness to share information (Zaheer medicine should concentrate on the efficiency and effective-
& Trkman, 2017), which is consistent with consumer trust ness of technology on improving physicians’ patient care and
theory that believes lack of consumer trust would hinder peo- service delivery rather than on the steps or procedures of the
ple’s communication in the online communities (Hoffman actual use of the technology (Hu et al., 1999).
et al., 1999). The theory of information sharing also recog- According to Marangunić and Granić (2015), TAM has
nizes the impact of privacy that reflects in the consumer trust experienced four major types of modifications in recent years,
theory with the emergence of data mining and data storage incorporating external predictors (e.g., technology anxiety,
that worsens the privacy protection on social media. Privacy prior usage, and experience, self-efficiency, and confidence
has been regarded as the most significant factor between these in technology), factors from other theories (e.g., expectations,
two theories (Razavi & Iverson, 2006). Besides, the ultimate user participation, risk, and trust), contextual factors (e.g., cul-
goal of these two theories is to help organizations to increase tural diversity and technology characteristics) and usage mea-
competitive competence, and from the consumer’s point of sures (e.g., attitudes towards technology, usage perception and
view, the premise of theories is to benefit humans in the digital actual usage of technology). Users’ anxiety, felt risks, trust
health domain through social media. The antecedent of reci- crisis, technology characteristics and their attitudes towards
procity can be reflected in the beneficence of responsible AI AI, in this case, might be valuable considerations when
while the increase of trust positively influences people’s men- users are assessing the responsibilities of AI in social media.
tal health. Therefore, integrating consumer trust theory and People felt anxious about the emergent technologies
information sharing theory assists us in better understand particularly when the technology interferes with their life
 Inf Syst Front

and there might be some uncertain risks such as the formulates the purpose of this investigation and the concep-
unconsciously exploited human rights and invasion of tion of the theme to be investigated, obtain a pre-knowledge of
privacy. Users are less likely to trust the technology and the subject matter, plan the design of the study, and conduct
pertinent changes. AI technology itself in social media is the interviews with a reflective approach to the knowledge
intangible and users cannot see, touch, and feel it, but they sought and the interpersonal relationship of the interview sit-
can experience the changes that occurred. Consequently, uation (Kvale, 2007). The participants selected in this research
TAM theory is of great importance in analyzing the research all have some working experience in sharing information on
question for this research. The authors extended the social media in China. They either promoted the medical ser-
examination of the issue through comprehensive vice or products on social media platforms such as Weibo,
consideration of the TAM and its associated modifications WeChat, and relevant health communities or answered pa-
to assess the performance of social media marketing. tients’ enquiries, monitoring public health on social media
pages of professional medical consultation software. AI tech-
2.9 Ethical Use of Data nology was deeply embedded in social media and people
working for them are enabled by AI capabilities to improve
The data is the core in the information era and the use of the their working performance.
majority of advanced technologies including AI. Burkhardt
et al. (2019) clarified that when AI has been quickly emerging 3.1 Epistemological Approaches to Responsible AI
as a new tool for CEOs to drive revenues and profitability,
CEOs should offer guidance to enable analytics teams to de- Qualitative analysis in academic marketing involves ap-
velop and use AI in an ethical way. Data acquisition should be proaches that match across the centre of the spectrum, extend-
aligned with stakeholders’ expectations for the use of their ing to the construction end (Hanson & Grimmer, 2007). The
data; dataset should reflect real-world populations rather than purpose of such research is to develop insights rather than
excluded data from minority groups; fairness should be con- measure, to explore rather than pin-down (ibid.). Interviews
sidered in the development process such as data selection, allow for a comprehensive investigation of human activities
feature selection, and model construction and monitoring; AI that can be aimed at practical, complex, and commercially
teams should use the simplest performance model and latest important issues such as consumer preferences
explicability techniques to meet different groups’ demands. (Gummesson, 2005). Due to the characteristics of this re-
Ethical use of data is closely intertwined with responsible AI search, the authors draw upon the interpretive methodologies
principles both of which are commonly accepted guidance. which aim to achieve substantive meaning and understanding
Ethical use of data, however, is seemingly superficial in of how and why questions regarding the phenomena under
actual applications of advanced technologies and requires investigation in the marketing, in managerial and consumer
fortified inspection. Cheng et al. (2021) recognized data dig- contexts (Carson et al., 2001). To gain validity and trustwor-
nity as an urgent issue that helps users to identify the benefits thiness in this qualitative research, the authors pay close atten-
and risks concerning their digital presence and personal data, tion to the careful use, interpretation, examination, and assess-
informs users of why and how their data will be used and ment of appropriate literature; careful justification of the qual-
allows users to negotiate the terms of using their data. itative research methodologies employed in this research, and
Ethical data becomes a key issue for the responsible develop- specifically the appropriateness, merits, and values; careful
ment of AI in social media and is particularly significant for structuring of interview analysis to ensure comprehensive
health industries. Each stage of collecting and processing AI and descriptive evaluation and assessment (ibid.).
data must be ethical. Data collectors and processors should
deliberate on the responsible AI principles, including fairness, 3.2 Sample Design
inclusiveness, liability and safety, transparency, privacy is-
sues, beneficence, non-maleficence, and autonomy. In-depth interviews were conducted online among partici-
pants from Chinese health industries, including one hospital,
one healthcare centre, and one medical centre, to collect pri-
3 Methodology mary data that reflect 25 social media executives’ and general
staff’s insights, working experience, and concerns towards
This paper performed a discovery-oriented research instru- responsible AI. (see Table 3) Clinicians from a Chinese hos-
ment, qualitative interviews to construct study-specific sets pital were included because they were dedicated to communi-
of questions that are open-ended in nature so the participants cating with patients through social media pages and answering
can contribute their insiders’ perspectives with little or no enquiries from them as part of their job duties. Social media
limitations imposed by more closed-ended questions executives and staff from a healthcare centre and a medical
(Chenail, 2011). Before the interview started, this paper company were interviewed because they were broadcasting
Inf Syst Front

Table 3 Profile of 25 participants

in the interviews Participants Number of participants Occupations of participants Industry sector

Participant A-G 7 Social media staff (general)-Clinician Hospital

Participant H-I 2 Social media executives Hospital
Participant J-M 4 Social media executives Healthcare center
Participant N-S 6 Social media staff (general)- Clinician Healthcare center
Participant T-U 2 Social media executives Medical corporation
Participant V-Y 4 Social media staff (general) Medical corporation

their service and product information on social media, as well media; and depth interviews with 25 social media practitioners
as monitoring and looking after the health situation of their that have related working experience of social media commu-
users frequently. Depth interviews focused on social media nities. The authors integrated three overlapped principles (i.e.,
activities relating to health information searching and gener- justice, explicability, accountability) and remained eight ma-
ating, the user experience of platforms, as well as their recog- jor principles (i.e., fairness, inclusiveness, reliability and safe-
nition of responsible AI efforts. Depth interviews of one to ty, transparency, privacy and security, beneficence,
two hours’ length were conducted with interviewees who non-maleficence, and autonomy). Companies have
were identified using theoretical sampling through peer intro- endeavoured to translate these principles into actionable prac-
ductions and snowballing techniques. They have provided a tice and sometimes fall short of dictating specific actions in
wealth of professional insights and anecdotal evidence practice, so a variety of solutions are required (Peters et al.,
supporting the face validity of our propositions. 2020). Qualitative studies in this paper are carried out among
social media marketing executives and staff in healthcare in-
dustries to examine the execution of responsible AI principles.
3.3 Proposition Development

The set of propositions shown in Table 4 connects the eight 3.4 Construction of Interview Schedule
responsible AI principles identified by Microsoft AI (2020)
and Floridi et al. (2018) to the social media platforms for the We constructed the interview items from the abovementioned
use of digital health (Dredze, 2012). The research propositions conceptual categories about the responsible AI principles and
are informed by three sources: current research on responsible social media. These items determined the main structure stems
AI principles and practices; current health research on social formulated for our interview schedule (see Table 5). Our

Table 4 Propositions of the organizational implications for responsible AI

No. Responsible AI Social media channels for Organizational propositions Moderator

principles digital health information

1 Fairness Discovering health topics. P1: The principle of fairness facilitates the Consumer trust.
Bio-surveillance performance of social media marketing Data quality.
2 Inclusiveness monitoring. P2: The principle of inclusiveness facilitates the P9: Trust and data quality moderate the
Sharing self-management performance of social media marketing performance of social media marketing
3 Reliability and information. P3: The principle of reliability and safety facilitates
safety the performance of social media marketing
4 Transparency P4: The principle of transparency facilitates the
performance of social media marketing
5 Privacy and P5: The principle of privacy and security facilitates
security the performance of social media marketing
6 Beneficence P6: The principle of beneficence facilitates the
performance of social media marketing
7 Non-maleficence P7: The principle of non-maleficence facilitates the
performance of social media marketing
8 Autonomy P8: The principle of autonomy facilitates the
performance of social media marketing

Sources: Peters et al., 2020; Sanches et al., 2019; Olson, 2018; D’Alfonso, 2020; Floridi, 2018; Morley et al., 2020; Shaban-Nejad et al., 2018;
Brownstein et al., 2009; Dredze, 2012; Hoffman et al., 1999
 Inf Syst Front

Table 5 Construction of Interview Schedule for Domain of Responsible AI principles to social media application

Categories of Item Scheduled question stem and probe Reference for category
responsible AI no. development
principles

Fairness 1 Can you share with me the act of fairness regarding the use of AI and social media Peters et al., 2020; D’Alfonso,
in your organization? Probe: How is the effectiveness? 2020.
Inclusiveness 2 What have your organization done for engaging social media users? Probe: How Sanches et al., 2019; Osatuyi,
did they react? 2013
Reliability and safety 3 How to ensure the reliability and safety of AI in your organization? Probe: Did it Olson, 2018; D’Alfonso, 2020.
work?
Transparency 4 What have your organization done for improving the openness of AI? Probe: How Olson, 2018; Peters et al., 2020;
did users react? D’Alfonso, 2020.
Privacy and security 5 How to ensure the privacy and security of AI in your organization? Probe: Did it D’Alfonso, 2020; Dredze, 2012;
work? Razavi & Iverson, 2006.
Beneficence 6 Do you think the AI in your organization has benefited the users? Probe: How to Floridi, 2018; Clarke, 2019;
balance the interests between organization and users? Razavi & Iverson, 2006.
Non-maleficence 7 How to ensure the non-maleficence of AI in your organization? Probe: Did it work? Floridi, 2018; D’Alfonso, 2020
Autonomy 8 How to ensure the autonomy of AI in your organization? Probe: Did it work? Morley et al., 2019

discussion is organized by responsible AI principles and starts affirmative action and extra support for a group (Peters
with developing propositions related to social media platforms et al., 2020). However, the research on whether the fairness
for digital health use. This paper also proceeded to discuss the of AI technology use on a variety of social media platforms
moderating role of trust and data quality, and ascertain remains limited. The authors have interviewed social media
the validity (i.e., whether an interview study investigat- practitioners and found some divergent insights into fairness.
ed what is intended to be investigated), reliability (i.e.,
how consistent the results are), and generalizability (i.e.,
how to generalize the findings of an interview study to Personalized notifications on our social networking sites
larger groups) of the interview findings (Kvale, 2007). will be based on AI’s learning from their psychology and
behavioral patterns, as well as AI’s understanding of each
audience. Consumers don’t feel unfair when using social
4 Analysis of Data and Discussion media but may feel extremely targeted. (Participant D)

4.1 Analysis of Social Media Marketing Towards Based on background information, keywords searched,
Fairness browsing records and pages, social media platforms sent proc-
essed, analyzed, and customized health messages, such as
AI systems should be responsibly designed, developed, and health ads or other health bloggers to audiences. A certain
deployed with appropriate safeguards such as procedural fair- degree of prejudice and stereotypes therefore emerges, but
ness (Lyons et al., 2021), and fair AI should not lead to dis- the degree may be reluctant from time to time, one user to
criminatory influences on humans associated with race, ethnic another, for example, the machine could infer that a single girl
origin, religion, gender, sexual orientation, disability or other was pregnant depending on the already-known age informa-
situations (Benjamins et al., 2019). Fairness is a complex eth- tion. People may think that data do not suffer from human bias
ical principle that is closely related to justice and equality, itself, but the truth is that all decisions are deeply impacted by
though they are quite the same as either (Peters et al., 2020). society and stereotypes (Benjamins et al., 2019). Hence, mar-
This concept captures the fair distribution of benefits, risks, keters have acquainted that consumers may feel annoyed and
and costs to all people irrespectively of social class, race, offended by the excessively targeted experience.
gender, or other forms of discrimination (Sanches et al.,
2019). In terms of protected targets, the fairness matrix is
divided into three categories, including individual fairness Our platform would introduce products mistakenly
which means each person is treated equally, and group fair- sometimes, just according to age and gender of con-
ness which implies that different groups such as women and sumers, therefore, for example, a single girl has received
men are treated equally, as well as subgroup fairness (Cheng messages of pregnancy test products because of AI’s
et al., 2021). The principle of fairness may contribute to misunderstanding. (Participant M)
Inf Syst Front

Companies have created complicated and intelligent hier- Proposition 1: The principle of fairness facilitates the
archical membership systems to earn the interests of different performance of social media marketing
categories of members through subsections. They offer
completely different consumers different products, services,
and information. The dark side of these customized offers 4.2 Analysis of Social Media Marketing Towards
might be injustice, discrimination, and harm. The marketing Inclusiveness
strategy of targeting does not involve unfair intention, but the
targeting strategy and relevant AI efforts by nature segment Inclusiveness means engaging poor consumers and producers
audiences into diverse classifications, which makes audiences in the development, production, and use of the AI ecosystem,
being treated differently and unfairly. AI has its systems of and helping them benefit from such technology. Inclusiveness
execution standards created by AI designers. can be achieved through active participation in the design and
innovation process, increased access of the poor to technolog-
ical services, and therefore an increase in the self-reliance of
We have a big data business, such as a user pricing the rural poor (Sanches et al., 2019). AI-powered companies
system, there will be some “price killers” (e.g., three tend to involve and engage a wide range of consumers in
people may search for different prices in the same place depth through technological advantages and innovation.
at the same time for a product). To some extent, these Respecting the characteristics of audiences with various back-
policies might be against the principle of fairness. ground can be priorities regarding this principle.
(Participant R)

These user’s hierarchy systems, such as drug price sys- Inclusiveness requires us to promote products and ser-
tem and membership system in the online communities vices based on the interests and standards of different
where there will be some “price killer policy” (e.g., three groups of people, rather than recommending high price
people have searched for different drug prices in the same drugs and treatments to low-income groups, which in-
place simultaneously). These policies are against the prin- creases our costs. (Participant L)
ciple of fairness. AI learned so much information from
mobile phones through machine learning, then marked Social media platforms usually require users to agree to
prices for each piece of information and sent different some privacy or authorization agreement before using them.
notifications to grouped audiences. Organizations intend These agreements also become obstacles for people to be in-
to strike a balance between the principle of fairness and cluded in the platforms particularly some privacy or authori-
the major objectives of profits. zation policies that do not fit in with users’ interests and
Overall, companies are more likely to promise fairness, but preferences.
the essence of companies is for-profit which cannot be
neglected. The value that companies want might not be con-
sistent with what consumers pursue, and therefore many con- If they do not need the app, they would reject the license
sumers remain doubtful towards companies’ promise for fair- agreement which set a huge negative impact on social
ness, which becomes a tough question for organizers. There media marketing. Although users are entitled the rights
might be some stereotypes towards technology and compa- to refuse, if they refuse, they cannot use the social media
nies. With the unified technique of fairness, the process of application. (Participant A)
evaluation and correction for fair AI would be easier
(Benjamins et al., 2019). A company was assumed to wear a This is a very common and contradictory phenomenon.
cap of ‘fairness’, but it tends to be difficult to define whether Additionally, consumers’ concerns about the profit-oriented
the judgment and activities are fair or not. Companies use AI essence of companies prevent consumers to raise trust in ben-
to make a profit by intelligent grading. According to some eficial technology. Overall, embracing inclusiveness on social
interviewees, they believe that AI is a complete tool and media provides more chances for companies to engage more
means for companies to make profits. Many behaviours of potential platforms and AI technology users. Data analysts
companies, including AI efforts, are more likely to be what and managers can pay much more attention to the features
companies fabricate for greater profits. The extreme pursuit of information (such as topic, embedded video, embedded
for fairness is more likely to damage the interests of the com- audio, and response count) shared on social media which in-
pany and negatively influence the effectiveness of social me- dicates the credibility of information, to better understand how
dia marketing. Hence, this finding is partly contradictory to to engage consumers (Osatuyi, 2013). Besides, companies
the principle of fairness. But still, fairness should be an im- might need to improve the efficiency of technology and avoid
portant principle for companies to follow. the deviation level of costs when engaging more consumers.
 Inf Syst Front

Proposition 2; The principle of inclusiveness facili- sold to others. Many fraudulent calls derive from such unreli-
tates the performance of social media marketing. ability and unsafety.

Our platform will let users tick boxes including privacy

4.3 Analysis of Social Media Marketing Towards agreements and licensing agreements, but to be honest,
Reliability and Safety not too many users would carefully look into details of
these agreements, unless they are worrying, or only
Doctors have been worried about the way that AI is when it may endanger their interests, or cause extra
being developed. Taking artificially intelligent bots costs. (Participant O)
Babylon Health as an example, it cannot conduct a rel-
atively rigorous overhaul of the development process In relatively fewer situations, social media users would pay
and some of the clinical advice was not vetted, then careful attention to these agreements. In most cases, they are
its effectiveness was exaggerated (Olson, 2018). The more likely to ignore these requirements, which in turn exerts
increase in perceived reliability and safety, and therefore pressure on platforms to improve the reliability and safety of
trust in the social media platforms backed by AI tech- social media platforms and AI-related technology. However,
nology may close the reinforcing loop as it triggers social media users are not innocent of the drawbacks of these
people to shift from their current health-seeking chan- advanced technologies. Some of them use some plug-ins to
nels more to the use of responsible AI-powered plat- block ads on their Google Chrome and mobile devices, so that
forms. Reliability and safety are intrinsically significant so-called unsafe information may be blocked. But it does
particularly when introducing AI to users. Responsible not mean the platform and technology become compre-
AI requires designers and engineers to consider that AI hensively safe. In a nutshell, the principle of reliability
is for human beings, therefore AI should abide by the and safety are more likely to facilitate the performance
ethics and interests of specific groups, not harming the of social media marketing by increasing consumers’ rec-
interests of others. In a nutshell, AI companies should ognition of technology.
be as risk-free or as controllable as possible. If AI tech-
nology is allowed to develop freely beyond the control Proposition 3: The principle of reliability and safety
of human beings, the problem will become extremely facilitates the performance of social media marketing.
complicated and troublesome.

4.4 Analysis of Social Media Marketing Towards

We have realized the concerns of reliability and safety Transparency
from users because the function of machine learning-
enabled social media platforms to remember users’ The transparency principle of responsible AI requires de-
search records and accordingly predict their preferences signers of AI systems to be open to users for example letting
and also show relevant results occasionally across dif- them know how the design process emerging and evolving.
ferent platforms which lead to some of the users feel AI systems should be clear to users, and fully consider the user
unreliable and unsafe. For instance, once they searched profile to accommodate them to the transparency level need-
for a health-related message on one platform, the other ed, particularly when using third-party AI technology and
platform would immediately jump back to the relevant more relevant for applications such as medical diagnosis
link or products, which made them felt much terrible. (Benjamins et al., 2019). When seeking a transparent algo-
(Participant G) rithm, an understandable explanation of how AI operates is
developed (Cheng et al., 2021). This may not only build trust
Particularly when they searched for an embarrassing ail- from consumers but also increase the quality of relevant infor-
ment and some awkward messages would continuously mation generated. Building AI for healthcare requires a further
emerge on diverse websites and applications in a certain peri- reconsideration of how a specific technology is designed.
od albeit companies acclaim that the tracking and predicting These rules should involve clinicians who can type a range
obligation of AI is to offer much useful information. Many of probabilities of symptoms into the systems (Olson, 2018).
mobile and computer users tend to use a lot of remembering Giving a brief introduction to AI might be needed for some
password functions on their mobile phones and browser. people, but if companies want to introduce the operation prin-
However, AI’s machine learning capability is so strong that ciple or mechanism behind AI, they need to be cautious when
makes consumers feel not much certain about whether their reaching a high level of secret or have already been involved
password and information are secure enough and not being in the level of other stakeholders’ interests. Proper
Inf Syst Front

introductions might be appropriate, but it needs to be classi- information. Secondly, companies or third-party service pro-
fied based on the level of information. viders may use cookies and some tracking technologies (i.e.,
Some can be introduced, but some had better be retained. pixels, beacons, mobile application identifiers, and Adobe
For example, if the intellectual property rights and core tech- Flash technology) to recognize users, improve their experi-
nologies that are related to interests or secrets, especially re- ence, increase security, serve to advertise, and measure the
gionally and nationally, are presented publicly, they will have use and effectiveness of services. Additionally, consumer’s
an extremely negative impact. navigation on the platforms could be tracked and being
targeted by certain third-party advertising companies. For ex-
ample, patient users might be integrated into practice manage-
Our designer could tell consumers how the principle is ment systems, making a referral, sending a prescription to a
but can’t tell them the specific algorithm logic and pro- pharmacy, or sending a test to a clinical laboratory, and au-
cess, which is a huge project that is impossible to un- thorize platforms or third parties to run statistic research on
cover. (Participant B) individual or aggregate trends.

After all, AI is a machine that helps to learn people’s

thoughts and behaviour which are authentic and essential in- Consumers do have the right to opt-out of tracking pol-
formation. The users of AI have certain rights to know. icies, but the reality is that the majority of consumers
might have ignored the rights and followed the system-
atic setting by default. Even though geographically lo-
We are worried about the possibilities of whether AI can calized information may lead to many privacy concerns,
be interpreted sufficiently up to the current development some users might not be fully informed that they might
level of AI technology. Provided that AI is sufficiently be served with location-enabled service. (Participant I)
open to the public, whether audiences can understand AI
is still uncertain owing to the different levels of cogni- Whether a company can establish good credibility and cor-
tive competence among audiences. It is more likely that porate reputation greatly depends on how well it can keep the
users may never think of getting to know the design and privacy of customers. In developed countries, the probability
operation of AI. (Participant Y) of being pushed related searching records is relatively small,
which situation is much better in developing countries. In
Moreover, consumers and AI owners, as well as designers, other words, consumer privacy protection systems in develop-
may be standing in two completely different parties, ing countries are even worse. Many social media users alleged
representing different positions, which implies that the com- that their privacy must not be well protected. In the informa-
panies’ position can be quite different from consumers. There tion and digital situation, it is very difficult to completely
can be conflicts of interest and incorporating some places that protect privacy unless people do not use any electronic prod-
can’t be understood either. Overall, a certain level of transpar- ucts. And unless people do not use computers and mobile
ency of AI-enabled social media could close the distance be- phones, as well as any other electronic devices, people can
tween inanimate technology and its users which may have a do protect our privacy. Many companies like Apple would
positive influence on the performance of social media have greatly leveraged information security as a selling point,
marketing. especially after their system has been upgraded or when a new
design for improving security developed. This fact shows that
Proposition 4: The principle of transparency facilitates there must be a severe problem with the security of informa-
the performance of social media marketing. tion.

Proposition 5: The principle of privacy and security

4.5 Analysis of Social Media Marketing Towards facilitates the performance of social media marketing.
Privacy and Security

The current stage in the information life cycle, the nature of 4.6 Analysis of Social Media Marketing Towards
trust between the owner and the receiver of information, and Beneficence
the dynamics of the group or community would play a strong
role in the user’s sharing attitude (Razavi & Iverson, 2006). The principle of creating AI technology that is beneficial to
Social media platforms have multiple specific policies for humanity has been expressed in different ways, ranging from
protecting consumers’ privacy and security. Firstly, on some prospering for mankind and the preservation of a good envi-
platforms, users can choose whether to disclose their personal ronment for future generations (Floridi, 2018). Beneficial AI
 Inf Syst Front

should be human-centric and at the service of society, as well AI should be executed under the principle of beneficence,
as produce tangible benefits for people (Benjamins et al., although it is very difficult to recognize the accurate influence
2019). For a specific social media and health sphere, the prin- it has on human beings, still, it might attract much investment
ciple of beneficence might be bringing more possibilities, op- into this industry to broaden the beneficence of AI and social
portunities, alternatives for the current and potential con- media. Perceived benefits of real-time information sharing
sumers’ health. The application of AI technology in the med- lead to overall perceived purchase and repurchase behaviour
ical field has indeed brought a lot of benefits, for example, mediated by customer orientation (Ghouri and Mani, 2019).
some minimally invasive surgery with a micro camera can
detect and photograph feedback for treatment. Another exam- Proposition 6: The principle of beneficence facilitates
ple is to predict what kind of disease a person will have the performance of social media marketing.
through gene sequencing, and then carry out some early
prevention.
4.7 Analysis of Social Media Marketing Towards
Non-maleficence
Those people with mental disorders usually experience
widespread human rights violations, discrimination, and Non-maleficence refers that the creation, design, and applica-
stigma as well as overwhelming economic, social, and tion of AI should be cautions against various negative conse-
treatment costs, but social media along with AI quences of overusing or misusing AI technology (Floridi,
access provides them with alternative treatment sugges- 2018). AI systems should in no way lead to negative effects
tions. Our efforts do contribute to solving some prob- on human rights (Benjamins et al., 2019). Along with the
lems in society. (Participant J) promising future of AI, various risks and related warnings
arise from their technological innovations, such as the recur-
However, AI technology is a computational tool, and sive self-improvement of AI and the threats of an AI arms
whether it can take effects would highly depend on the orig- race. (ibid.) Thanks to the wide use of AI in the mobile internet
inal intention of organizers. Despite these efforts that social industry, which makes people feel that everything is faster and
media and AI practitioners have made so far, social media more convenient in today’s life, the sense of boundary has
users still hold doubts and concerns towards the beneficence become more unclear than before so the balance between life
objective of AI technology. Some businesses of non-profit and work as well as social life has been broken.
online services, which aim to collect data and use AI for public
welfare, eventually make these data are in the hands of some
private companies because many non-profit organizations Some of our users stated that AI in social media has
outsource a lot of business to private companies. The benefits been overused. They complained about the decreasing
of AI have not been admitted widely, which might owe to the time for rest and less private space since they need to be
whole economic and societal development of the society that ready for work and socialize anytime and anywhere.
propels the visibly beneficial AI. Still many globally social (Participant C)
problems, such as the gap between the rich and the poor, have
always existed and are not being relieved much, because many The overuse of AI has made people living in overwhelming
poor areas still have no access to the Internet. information conditions. People complain about the overuse of
AI because they have received overloaded messages every
day and most of them are useless and even annoying and
So far what AI has contributed seems limited, but AI can upset. Hence, the opposite of convenience and fast is that, to
actually have more advantages than disadvantages when some extent, AI harms people’s mental health and balanced
used properly. (Participant N) life. In a nutshell, there is still a way for organizations to
relieve the negative effects led by fast-paced life, and simul-
AI can become a promoter, not a chaos creator. AI is not taneously to make everything more convenient, faster, and
created for the benefit of mankind and therefore its usage efficient.
depends on those who have the right to operate and use it.
What companies need is to maximize the organizational inter-
ests, not to maximize the effectiveness of AI without consid- We recognized that AI has been misused to some extent.
ering the sake of companies. The interviewees commonly ad- Its original goal was to benefit mankind and promote
mitted the restricted contributions that AI had on humans, but human development, rather than to do AI for the bene-
they were not sure that the emergence and development of AI fits of the power and money of those who have acquired
will necessarily promote the development of human society. it. But the truth is that AI has been used for the sake of
Inf Syst Front

companies regardless of the obstructive effects it may elderly and the children, who have limited cognitive
have on people. (Participant D) abilities and recognizing true from false, is where AI
should entitle them more autonomy to help make deci-
For example, many social network software, such as sions, rather than lead them in the wrong direction or go
WeChat pay, Alipay, and bank card binding function, will astray. But for young and middle-aged people, the pro-
form a lot of big data reservoirs to help analyze the user’s portion of AI interruption in the consumer decision pro-
information. These platforms intend to build a huge network cess can be smaller. This is hard for machines based on
for you to analyze and control. In principle, these behaviours the existing technology. (Participant F)
should bring as much convenience to your life as possible, but
because its system is getting more complicated and large, In a nutshell, a different group of users and events require
people’s lives are becoming more inconvenient. To improve to be allocated to determine the degree of AI participation,
the non-maleficence of AI is more likely to facilitate the per- while AI is intelligent, it should make different efforts for
formance of social media marketing unless companies pay different people. Furthermore, for two adults of the same
much attention to the natural needs of humans such as much age, they may need a different level of AI interruption prefer-
quiet rest time. ence. For some people, AI are required to entitle more rights
and choices to help users make choices; for another user, they
Proposition 7: The principle of non-maleficence facil- may desire much dominant power and only need basic ser-
itates the performance of social media marketing. vices or suggestions from AI. Organizations claimed another
issue is that the range of permissions is difficult to define and
may need to be evaluated by specific design theories and
4.8 Analysis of Social Media Marketing Towards systems. For example, providing three to four suggestions
Autonomy may be considered to be highly autonomous, but more than
a dozen suggestions are not that autonomous.
AI must protect and enhance users’ autonomy and abil- To ensure the autonomy of social media users requires
ities to make decisions and choose between alternatives. companies to segment people into different groups which
Users should also be able to make informed autono- might have conflicts with the excessive targeted concerns of
mous decisions regarding AI systems. If not, users people. Unless this issue has been addressed, the proper level
may feel that their decisions are being curtailed by the of autonomy can both facilitate the performance of social me-
systems that they do not understand, and it is very un- dia marketing and ensure recognition of the company. More
likely that these systems will satisfy social acceptability, importantly, humans must be empowered with greater rights
regardless of social preference that should be the goal to monitor and intervene in AI if necessary. This consideration
of a truly ethically designed AI (Morley et al., 2020). A derives from predictions of AI as superintelligence in the fu-
trustworthy AI can be more socially acceptable (Taylor ture.
et al., 2018).
Proposition 8: The principle of autonomy facilitates the
performance of social media marketing.
We admitted that AI has deprived some of the users’
rights for general decisions in daily life. Users may hope 4.9 Consumer Trust and Data Quality as Moderators
to keep more of their rights. (Participant M)
Throughout the interviews with 25 social media practitioners,
Additionally, AI technology has relative stereotypes about this paper concluded two important moderators they have
each audience. It can help them to solve problems, but these mentioned as key drivers for the principles of responsible
are all based on its restricted machine learning and under- AI. They claimed that AI should build trustworthiness and
standing of the audience. However, human beings are much quality of data. The basic premise that consumers can admit
more subjective and emotional, so there are many aspects that AI is that they trust it (Ye, 2010). Particularly when AI is
AI can’t sufficiently understand. So far AI is not so accurate, helping people screen information and make decisions, the
so social media users would have to take some risks to follow quality of the information can help users identify whether to
its suggestions.” accept AI’s suggestions or not. For example, if a consumer has
searched for an anorectal disease, all of his/her search engines
and social platforms will pop up all kinds of related advertise-
Some people have different requirements towards au- ments, many of which are useless. This negative response
tonomy, ranging from diverse age generations, educa- hindered consumer’s level of use and trust towards AI on
tional backgrounds, and cultural differences. For the social media. So how to make AI technology more quickly
 Inf Syst Front

and accurately find the most suitable page for him, or the 4.10 The Exploration of a Conceptual Model on
solutions, is quite significant for consumers to build trust Responsible AI
and reliability towards AI.
Each principle of responsible AI and its corresponding conse-
quences are shown in Fig. 2. It has been found in this research
We know that AI should recommend information accu- that the obedience of each responsible AI principle would
rately and objectively. (Participant K) engender significantly positive results for organizations.
Firstly, achieving fairness is difficult, but organizations are
However, a small group of people felt panicked when a required to strike a balance between profits and equity and
dispensable platform sent accurate messages to them frequent- also pay closer attention to less privileged groups. This would
ly, because it implied that AI understood them exceptionally increase users’ willingness to engage in the technology and
well and beyond the restrictions of norms. Hence, it tends to associated social media activities. Secondly, organizations are
be critically essential when companies leverage AI to commu- encouraged to involve more diverse audiences, respect their
nicate with consumers without disturbing and offending them. distinctive characteristics, and encourage active participation.
To build long-lasting trust between AI and human beings, the This is in line with the born nature of responsible AI which
key is to address AI’s indifferent behaviour (Cheng et al., cares about the less privileged residents and aims to help them
2021). survive and develop from the technology innovation. To
Proposition 9: Consumer trust and data quality mod- broaden the potential audiences also entitles the organization’s
erate the performance of social media marketing. opportunities to better understand one group and fulfill their
demands. Inclusiveness requires organizations to focus on the
The above analysis and discussion on each principle and characteristics of different groups of people, understand and
activities of responsible AI led to the development of the fol- respect them, which seems difficult, but these minor groups
lowing conceptual model (see Fig. 1) which demonstrated the would reward the efforts. Thirdly, reliable information in so-
research questions and relevant factors that influence social cial media for health care is much more important than in any
media marketing and digital health from both the organiza- other field since it is highly linked with the safety and health of
tional level and consumer level. According to Cheng et al. each patient. Besides, data security cannot be ignored and
(2021), three concrete objectives of trust are fairness, trans- should be listed as priorities for organizations at any time.
parency, and safety. Data quality is what can be addressed at The usefulness of technology in this research is more about
the organizational level while consumer trust is generated the authority of the information; reliability and safety of the
from the consumer level but rooted in the people’s acceptance data are to be the most fundamental ingredients. With a higher
of AI and influenced by many contextual factors. The inter- level of reliability and safety of AI technology, users are more
play effects between these two moderators can help to likely to perceive the usefulness of the technology to be will-
strengthen the effectiveness of appliance of responsible AI ing to participate in it. Fourthly, responsible AI ought to be
principles. relatively transparent. This is not only because of the com-
plexity of health information but also a thoughtful

$ $
6RFLDO
5HVSRQVLEOH 5HVSRQVLEOH
PHGLD
$,SULQFLSOHV $,DFWLYLWLHV
PDUNHWLQJ
2UJDQL]DWLRQDO
OHYHO

$
'DWDTXDOLW\ 
%

&RQVXPHU
'LJLWDOKHDOWK OHYHO
&RQVXPHUWUXVW

Fig. 1 A conceptual model of the responsible AI effects. *A1: responsible AI appliance in social media contributes to public digital
Responsible AI principles guide responsible AI activities about how to health, which resulted in the individual’s improvement in health
plan, design, and implement AI more responsibly. A2: Responsible AI conditions. B1: The level of data quality and consumer trust moderate
activities are tightly integrated with social media marketing, which aims the implementation of responsible AI principles in practices.
to address the ethical issues confronted with these activities. A3: The
Inf Syst Front

Fig. 2 Responsible AI principles

and corresponding consequences

consideration to maintain the citizen’s rights to be informed. enhance the ease of AI use, engender a higher level of user
Users on the platforms are entitled to comprehend how their acceptance and intention to use the technology. Hence, the
data is used, where to go, and who is to handle it to avoid any TAM model also contributed to the theoretical construction
unethical or illegal use of their private information. Fifthly, the of this paper. This paper leveraged the elements of the TAM
issue of protecting privacy in the digital world is more than a model to assess the effectiveness of responsible AI use. For
platitude and was regarded to be more severe than ever, as AI example, social media users would be more likely to ac-
has leveraged some advanced skills to mimic people and even cept the technology if it has sufficient ease of use and
intelligently act like mankind. Seeking strict privacy protec- perceived usefulness. Additionally, a higher level of
tion could ensure a high level of good credibility and corpo- consumer trust and information quality would strengthen
rate reputation which are worthy of equity for the sustainable the consumers’ confidence in a specific technology,
development of the organization. Sixthly, AI-enabled social resulting in more user traffic and participation in the
media marketing facilitates health care for a specific group of social media platforms.
patients, although many people are concerned about the be-
neficence goals of AI. Organizations can make more efforts in
improving the well-being of humans and better preserving the 5 Theoretical and Practical Implications
environment. If users’ realization of the usefulness of respon-
sible AI for health management increased, the acceptance of This research proposes and empirically tests a model of social
the technology would soon improve. Seventhly, overuse and media marketing reflecting responsible AI relationships guid-
misuse of AI are detrimental and what’s worse, the situations ed by specific principles, with moderating factors that are
might happen in many timing periods of AI design and consumer trust and data accuracy. Whoever uses AI can eth-
implementations which will wipe out hardly constructed user ically use it, since how well ethics are treated will ultimately
trust. Overloaded information has led to complicated and decide how much people will embrace technology in the fu-
messy human life; over-processed information has worsened ture. AI practitioners and researchers should be cautious about
the invasion of privacy. The hieratical level of digital health ethical issues. The more educated a criminal is, the more evil
should not be neglected like destroyed mental health and bal- they will be. If AI becomes more and more intelligent, and
anced life. Therefore, maintaining the non-maleficence of AI people who create and use AI don’t have enough ethical
particularly when facing health issues is of much concern. knowledge, the AI-empowered world would be exceedingly
Finally, the design and implementation of responsible AI troublesome. This paper has provided some theoretical and
should respect the subjective initiatives of humans rather than practical contributions for marketers and AI developers to
deprive them of an increasing level of autonomy. Users’ ac- consider ethical issues.
ceptability and intention to use the technology would increase
by ensuring a certain level of their autonomy to choose, decide 5.1 Theoretical Implications
and refuse.
The results extended the responsible AI theory by forming This paper proposed a set of eight significant responsible AI
a new set of responsible AI principles, and further validate principles framework and conducted a systematic discussion.
their significances in indicating positive results for digital Then the study suggested the usefulness of the framework in
health practitioners when they are implementing the social implementing activities in social media when applied in the
media campaign. The appliance of responsible AI may greatly digital health industry. Additionally, the interviews also
 Inf Syst Front

provide empirical evidence for the development of responsi- organizations, they can perceive information sharing as a loss
ble AI principles. of power so they would not be willing to share information
In the framework, data quality and consumer trust were (Zaheer & Trkman, 2017). These concerns contradicted what
regarded as essential elements that moderated the principles responsible AI wishes organizations to do. For the theory of
and activities. Apart from that, this paper also contributed to information sharing, privacy is a key issue to remain the
the responsible AI theories and social media marketing theo- long-term competitiveness of organizations (Razavi &
ries from combined organizational perspectives and consumer Iverson, 2006), which is consistent with responsible AI
perspectives. It implies that if an organization seeks to sustain principles.
a long-term relationship with consumers, it should abide by
the eight principles and concentrate on building trust,
collecting, and ethically processing the data. 5.2 Practical Implications
The information-sharing theory is applicable in many dis-
ciplines of social science such as supply chain and information Overloaded information, false reporting, lack of signal preci-
systems (Wu et al., 2014; Jarvenpaa & Staples, 2000). This sion, and exposure to external forces such as media interest
paper combined consumer trust theory (Lou & Yuan 2019) to may restrict the realization of their potential for public health
add new insights into social media and responsible AI princi- practice and clinical decision-making (Brownstein et al.,
ples to the research of the information sharing theory. The 2009). The current study argues that for digital health to be
theory of information-sharing offers an understanding of the effective, it should not ignore the social media impact coupled
variables that enable and constrain information exchange with responsible AI principles by accelerating the
among individuals (Zaheer & Trkman, 2017). The essence scrutinization of individuals’ feelings, responses, and insights
of social media platforms is important channels for people on social media. The results have answered the aforemen-
and organizations to share information. Hence, the appliance tioned research questions, as the results showed that inter-
of AI technology may boost or constrain the information ex- viewers are expecting AI technology and companies can pri-
change amount and information quality which depends on the oritize consumers’ interests rather than profits, abide by ethi-
ethical issues that organizations need to obey. cal principles, and be in line with consumer preference and
Trust is the most crucial antecedent and motivation for acceptance adequately when they undertake activities. The
consumers’ willingness to share (Zaheer & Trkman, 2017). efforts that a company makes for responsible AI should also
The moderator in this paper, the level of consumer trust would correspond to the interests of consumers and their individual-
influence an individual’s willingness to share information on ized demands and situations. Our study has proposed some
social media. It is believed that investing in social values new comprehension of what responsible AI should like and
based on trust, mutuality, and respect could enable what aspects are consumers concerned about. Companies can
long-term organizational benefits such as corporate create different choices of AI systems to engage consumers
well-being and innovativeness (Widén-Wulff & Ginman, from different backgrounds in catering to their demands; com-
2004). Good quality information, specifically reliable and ac- panies can construct a diverse level of openness for the AI
curate information sharing, is not possible on social media design process and relevant knowledge for customized desires
without trust from users (Kwon & Suh, 2005). The reciprocity of consumers; the design and intentions of responsible AI
consideration from information sharing theory supports the should prioritize consumers’ interests.
beneficence principle of responsible AI, which indicates that The principles of fairness, inclusiveness, reliability, and
when people feel potential benefits of specific behaviour, they safety, transparency, privacy and security, beneficence,
are more likely to get involved in the social media activities of non-maleficence as well as autonomy play an indispensable
organizations. role for companies applying AI technology responsibly.
Information sharing theory requires AI developers to create Precisely, the principle of transparency and autonomy are rel-
user-friendly products that fit in with their acceptance and atively personalized options for different consumers.
behaviour (Kim & Lee, 2006), which supports the transparen- Reliability and safety can be enhanced by upgrading data
cy principle of responsible AI that calls for customized deliv- quality and raising trust. Beneficence and non-maleficence
ery of openness. Information sharing theory includes task in- are more likely to be rooted in the infant intentions of compa-
terdependence as an antecedent that when people feel sharing nies, which are extremely basic and essential. Fairness and
information on social media is a social good, rather than per- inclusiveness tend to be related to prejudice and stereotype
sonally costly or unpleasant, it tends to be beneficial to the which may lead to consumer dissatisfaction and turndown to
organizations in the long run (Constant et al., 1994). From the AI technology. What interviewees have mentioned most is
theory of information sharing, power is not a significant ante- privacy. The innovation of technology improved part of pri-
cedent of willingness to share; high power might engage the vacy but also eroded the level of privacy. Overall, this re-
user to share information albeit unwillingly. For search offers important practical insights into ethics for AI
Inf Syst Front

companies and designers, which fills up with the scarce of Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing, adap-
theoretical knowledge in this field. tation, distribution and reproduction in any medium or format, as long as
you give appropriate credit to the original author(s) and the source, pro-
vide a link to the Creative Commons licence, and indicate if changes were
made. The images or other third party material in this article are included
6 Limitations and Suggestions for Future in the article's Creative Commons licence, unless indicated otherwise in a
Research credit line to the material. If material is not included in the article's
Creative Commons licence and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain
Responsible AI is still in its nascent stages and there are still permission directly from the copyright holder. To view a copy of this
few commonly accepted standards (Taulli, 2021; Fosso licence, visit http://creativecommons.org/licenses/by/4.0/.
Wamba & Queiroz, 2021). As with most empirical studies,
this research is not without its limitations although every effort
was made to minimize them. First, this research mostly relied References
on collecting data from single respondents – i.e., social media
executives and general staff. Although, this paper ensured that Ahmed, W., Bath, P. A., Sbaffi, L., & Demartini, G. (2019). Novel in-
these respondents are highly experienced employees who sights into views towards H1N1 during the 2009 Pandemic: a the-
matic analysis of Twitter data. Health Information & Libraries
have years of experience working in this field and carry a Journal,36(1), 60–72
wealth of knowledge, however, the fact that data was mostly Arrieta, A. B., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A.,
collected from a single source may have had an impact on its Tabik, S., Barbado, A., & Herrera, F. (2020). Explainable
richness as multiple participants would have provided more Artificial Intelligence (XAI): Concepts, taxonomies, opportu-
nities and challenges toward responsible AI. Information
insights. For instance, data can be collected from various
Fusion,58, 82–115
stakeholders beyond organizational employees (e.g., cus- BAAI. Artificial Intelligence for Children: Beijing Principles. Beijing
tomers, suppliers, etc.) to provide new insights. The paper Academy of Artificial Intelligence. https://www.baai.ac.cn/ai-for-
would, therefore, recommend that scholars conducting future children.html. Accessed 29 Sept 2020
research in this topic area focus on collecting data from mul- Benjamins, R. (2020). Towards organizational guidelines for the respon-
sible use of AI.https://arxiv.org/pdf/2001.09758. Accessed 30
tiple participants. Second, and in a similar vein, the data col- Sept 2020
lected in this research was mostly gathered through qualita- Benjamins, R., Barbado, A., & Sierra, D. (2019). Responsible AI by
tive, semi-structured interviews which are aligned to the ex- design in practice. https://arxiv.org/pdf/1909.12838. Accessed 30
ploratory nature of this research. Employing quantitative and Sept 2020
multi-method research would not only make this research Booth, R. G., Allen, B. N., Jenkyn, K. M. B., Li, L., & Shariff, S. Z.
(2018). Youth mental health services utilization rates after a large-
more robust but will also open new this research to new meth- scale social media campaign: population-based interrupted time-
odological directions. For example, for future research in re- series analysis. JMIR Mental Health,5(2), 1–15
sponsible AI, this work recommends a multilevel study with Briand, A., Almeida, H., & Meurs, M. J. (2018). Analysis of social media
clinicians and AI designers. Third, this research primarily in- posts for early detection of mental health conditions. Advances in
Artificial Intelligence, 133–143. https://doi.org/10.1007/978-3-319-
volved examining the application of responsible AI on social 89656-4_11
media in digital health marketing. Although there is potential Brownstein, J., Clark, S., Freifeld, C., Lawrence, C., & Madoff. (2009).
in our findings to be generalized to other industry sectors, Digital disease detection — Harnessing the web for public health
there is no doubt, that the results are more applicable and surveillance. The New England Journal of Medicine,360(21), 2153–
2157. https://doi.org/10.1056/NEJMp0900702
useful for the healthcare sector. However, there is potential
Burkhardt, R., Hohn, N., & Wigley, C. (2019). Leading your organization
for this research to be extended to other areas of service sec- to responsible AI.McKinsey Analytics,1–8
tors such as the use of socially responsible AI in banking and Carson, D., Gilmore, A., Perry, C., & Gronhaug, K. (2001). Qualitative
fraud, and subsequently, future researchers can also look into marketing research. Sage
examining the application of socially responsible AI in other, Chenail, R. J. (2011). Interviewing the investigator: Strategies for ad-
dressing instrumentation and researcher bias concerns in qualitative
more traditional industry sectors. Finally, there are some the- research. Qualitative Report,16(1), 255–262
oretical areas for further research based on this paper. For Cheng, L., Varshney, K. R., & Liu, H. (2021). Socially responsible AI
instance, to construct the mechanism for governance and con- algorithms: issues, purposes, and challenges. https://arxiv.org/abs/
trol of privacy and build trust in social media marketing would 2101.02032. Accessed 1 Nov 2020
be recommendable in the field. In summary, for the most part, Clarke, R. (2019). Principles for rsponsible AI. https://tech.humanrights.
gov.au/sites/default/files/inline-files/4A%20-%20Roger%20Clarke.
the limitations of this research are those that are common to pdf. Accessed 1 Nov 2020
exploratory, qualitative studies. Nonetheless, given the initial Coiera, E. (2015). Guide to health informatics. CRC Press
stage of growth of this research area, this paper is intrigued to Constant, D., Kiesler, S., & Sproull, L. (1994). What’s mine is ours, or is
find out how future research in this area moves forward. it? A study of attitudes about information sharing. Information
Systems Research,5(4), 400–421
 Inf Syst Front

Contractor, D., McDuff, D., Haines, J., Lee, J., Hines, C., & Hecht, B. Hu, P. J., Chau, P. Y., Sheng, O. R. L., & Tam, K. Y. (1999). Examining
(2020). Behavioral use licensing for responsible AI. https://arxiv. the technology acceptance model using physician acceptance of
org/abs/2011.03116. Accessed 1 Nov 2020 telemedicine technology. Journal of Management Information
D’Alfonso, S. (2020). AI in mental health. Current Opinion in Systems,16(2), 91–112
Psychology,36, 112–117 Jarvenpaa, S. L., & Staples, D. S. (2000). The use of collaborative elec-
Davis, F. D. (1985). A technology acceptance model for empirically test- tronic media for information sharing: an exploratory study of deter-
ing new end-user information systems: Theory and results (Doctoral minants. The Journal of Strategic Information Systems,9(2–3), 129–
dissertation, Massachusetts Institute of Technology) 154
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user Kapoor, K. K., Tamilmani, K., Rana, N. P., Patil, P., Dwivedi, Y. K., &
acceptance of information technology. MIS Quarterly, 319–340 Nerur, S. (2018). Advances in social media research: Past, present
Denecke, K., Gabarron, E., Grainger, R., Konstantinidis, S. T., Lau, A., and future. Information Systems Frontiers,20(3), 531–558
Rivera-Romero, O. … Merolli, M. (2019). Artificial intelligence for Kim, S., & Lee, H. (2006). The impact of organizational context and
participatory health: applications, impact, and future implications. information technology on employee knowledge-sharing capabili-
Yearbook of Medical Informatics, 28(1), 165–173 ties. Public Administration Review,66(3), 370–385
Doan, S., Ngo, Q. H., Kawazoe, A., & Collier, N. (2019). Global health King, W. R., & He, J. (2006). A meta-analysis of the technology accep-
monitor: A web-based system for detecting and mapping infectious tance model. Information & Management,43(6), 740–755
diseases. https://arxiv.org/abs/1911.09735. Accessed 31 Oct 2020 Kvale, S. (2007). Planning an interview study. Doing Interviews,1, 34–51
Dredze, M. (2012). How social media will change public health. IEEE Kwon, I. W. G., & Suh, T. (2005). Trust, commitment and relationships
Intelligent Systems,27(4), 81–84 in supply chain management: a path analysis. Supply Chain
Eitel-Porter, R. (2021). Beyond the promise: implementing ethical AI. AI Management,10(1), 26–33. https://doi.org/10.1108/
and Ethics,1(1), 73–80 13598540510578351
European Group on Ethics in Science and New Technologies. Statement Lee, Y., Kozar, K. A., & Larsen, K. R. (2003). The technology accep-
on artificial intelligence, robotics and ‘autonomous’ systems. tance model: Past, present, and future. Communications of the
https://ec.europa.eu/info/news/ethics-artificial-intelligence- Association for Information Systems,12(1), 50
statement-ege-released-2018-apr-24_en. Accessed 31 Oct 2020 Lima, G., & Cha, M. (2020). Responsible AI and its stakeholders. https://
Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., arxiv.org/abs/2004.11434. Accessed 10 Nov 2020
Dignum, V., & Vayena, E. (2018). AI4people-an ethical framework
Lou, C., & Yuan, S. (2019). Influencer marketing: how message value
for a good AI society: opportunities, risks, principles, and recom-
and credibility affect consumer trust of branded content on social
mendations. Minds and Machines,28(4), 689–707
media. Journal of Interactive Advertising,19(1), 58–73
Fosso Wamba, S., Queiroz, M.M. (2021). Responsible Artificial
Lyons, H., Velloso, E., & Miller, T. (2021). Fair and Responsible AI: A
Intelligence as a Secret Ingredient for Digital Health: Bibliometric
focus on the ability to contest.https://arxiv.org/abs/2102.10787.
Analysis, Insights, and Research Directions. Inf Syst Front. https://
Accessed 10 Nov 2020
doi.org/10.1007/s10796-021-10142-8
Marangunić, N., & Granić, A. (2015). Technology acceptance model: a
Ghallab, M. (2019). Responsible AI: requirements and challenges. AI
literature review from 1986 to 2013. Universal Access in the
Perspectives,1(1), 1–7
Information Society,14(1), 81–95
Ghebreyesus, T. A. (2019). The WHO special initiative for mental health
(2019–2023): Universal health coverage for mental health. World Mcclellan, C., Ali, M. M., Mutter, R., Kroutil, L., & Landwehr, J. (2017).
Health Organization. https://apps.who.int/iris/bitstream/handle/ Using social media to monitor mental health discussions - evidence
10665/310981/WHO-MSD-19.1eng.pdf?sequence=1&isAllowed= from Twitter. Journal of the American Medical Informatics
y. Accessed 30 Oct 2020 Association,24, 496–502
Ghouri, A. M., & Mani, V. (2019). Role of real-time information-sharing Meske, C., Bunde, E., Schneider, J., & Gersch, M. (2020). Explainable
through SaaS: An industry 4.0 perspective. International Journal of artificial intelligence: objectives, stakeholders, and future research
Information Management,49, 301–315 opportunities. Information Systems Management, 1–11
Gooding, P. (2019). Mapping the rise of digital mental health technolo- Microsoft, A. I. (2020). Responsible AI. https://www.microsoft.com/en-
gies: Emerging issues for law and society. International Journal of us/ai/responsible-ai?activetab=pivot1:primaryr6. Accessed 4
Law and Psychiatry,67, 1–11 Oct 2020
Gummesson, E. (2005). Qualitative research in marketing. European Mirbabaie, M., Ehnis, C., Stieglitz, S., Bunker, D., & Rose, T. (2020).
Journal of Marketing,39(3–4), 309–327. https://doi.org/10.1108/ Digital nudging in social media disaster communication.
03090560510581791 Information Systems Frontiers, 1–17
Gupta, A., Lanteigne, C., & Heath, V. (2020). Report prepared by the Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020). From what to
Montreal AI Ethics Institute (MAIEI) for Publication Norms for how: an initial review of publicly available AI ethics tools, methods
Responsible AI by Partnership on AI. https://arxiv.org/abs/2009. and research to translate principles into practices. Science and
07262. Accessed 4 Oct 2020 Engineering Ethics,26(4), 2141–2168
Hanson, D., & Grimmer, M. (2007). The mix of qualitative and quanti- Olson, P. (2018). This health startup won big government deals–but
tative research in major marketing journals, 1993-2002. European inside, doctors flagged problems. Forbes. https://www.forbes.com/
Journal of Marketing,41(1/2), 58–70. https://doi.org/10.1108/ sites/parmyolson/2018/12/17/this-health-startup-won-big-
03090560710718111 government-dealsbut-inside-doctors-flaggedproblems/?sh=
Hatala, J. P., & George Lutta, J. (2009). Managing information sharing 774efad0eabb. Accessed 30 Oct 2020
within an organizational setting: A social network perspective. Osatuyi, B. (2013). Information sharing on social media sites. Computers
Performance Improvement Quarterly,21(4), 5–33 in Human Behavior,29(6), 2622–2631
Hoffman, D. L., Novak, T. P., & Peralta, M. (1999). Building consumer Peters, D., Vold, K., Robinson, D., & Calvo, R. A. (2020). Responsible
trust online. Communications of the ACM,42(4), 80–85 AI—two frameworks for ethical design practice. IEEE Transactions
Holden, R. J., & Karsh, B. T. (2010). The technology acceptance model: on Technology and Society,1(1), 34–47
its past and its future in health care. Journal of Biomedical Puhakka, R., Ollila, S., Valve, R., & Sinkkonen, A. (2019). Consumer
Informatics,43(1), 159–172. https://doi.org/10.1016/j.jbi.2009.07. trust in a health-enhancinginnovation–comparisons between
002 Finland, Germany, and the United Kingdom. Journal of
Inf Syst Front

International Consumer Marketing,31(2), 162–176. https://doi.org/ media posts about an emerging suicide game. Journal of
10.1080/08961530.2018.1498757 Adolescent Health,65(1), 94–100
Rai, A., Patnayakuni, R., & Seth, N. (2006). Firm performance impacts of Szajna, B. (1996). Empirical evaluation of the revised technology accep-
digitally enabled supply chain integration capabilities. MIS tance model. Management Science,42(1), 85–92
Quarterly,30(2), 225–246. https://doi.org/10.2307/25148729 Taulli, T. (2021). Responsible AI: ethics and transparency. In Taulli, T.
Rauniar, R., Rawski, G., Yang, J., & Johnson, B. (2014). Technology (Eds.), Implementing AI Systems: Transform Your Business in 6
acceptance model (TAM) and social media usage: an empirical Steps. 174–199
study on Facebook. Journal of Enterprise Information Taylor, S., Pickering, B., Boniface, M., Anderson, M., Danks, D.,
Management,27(1), 6–30. https://doi.org/10.1108/JEIM-04-2012- Følstad, A. … Woollard, F. (2018). Responsible AI–Key themes,
0011 concerns & recommendations for European research and
Razavi, M. N., & Iverson, L. (2006). A grounded theory of information innovation. Next Generation Internet. https://www.ngi.eu/wp-
sharing behavior in a personal learning space. In Proceedings of the content/uploads/sites/48/2018/07/Responsible-AI-Consultation-
2006 20th anniversary conference on Computer supported cooper- Public-Recommendations-V1.0.pdf. Accessed 20 Nov 2020
ative work. 459–468 Trocin, C., Mikalef, P., Papamitsiou, Z et al. (2021). Responsible AI for
Rocha, H. M., Savatt, J. M., Riggs, E. R., Wagner, J. K., Faucett, W. A., Digital Health: a Synthesis and a Research Agenda. Inf Syst Front.
& Martin, C. L. (2018). Incorporating social media into your support https://doi.org/10.1007/s10796-021-10146-4
tool box: points to consider from genetics-based communities.
Tutubalina, E., & Nikolenko, S. (2018). Exploring convolutional neural
Journal of Genetic Counseling,27(2), 470–480. https://doi.org/10.
networks and topic models for user profiling from drug reviews.
1007/s10897-017-0170-z
Multimedia Tools and Applications,77(4), 4791–4809
Rousidis, D., Koukaras, P., & Tjortjis, C. (2020). Social media prediction:
a literature review. Multimedia Tools and Applications,79(9), 6279– Wahl, B., Cossy-Gantner, A., Germann, S., & Schwalbe, N. R. (2018).
6311 Artificial intelligence (AI) and global health: how can AI contribute
Sanches, P., Janson, A., Karpashevich, P., Nadal, C., Qu, C., Daudén to health in resource-poor settings? BMJ Global Health,3(4),
Roquet, C. … Sas, C. (2019). HCI and Affective Health: Taking e000798–e000798. https://doi.org/10.1136/bmjgh-2018-000798
stock of a decade of studies and charting future research directions. Wearn, O. R., Freeman, R., & Jacoby, D. M. (2019). Responsible AI for
In Proceedings of the 2019 CHI Conference on Human Factors in conservation. Nature Machine Intelligence,1(2), 72–73
Computing Systems. 1–17 Widén-Wulff, G., & Ginman, M. (2004). Explaining knowledge sharing
Scantamburlo, T., Cortés, A., & Schacht, M. (2020). Progressing to- in organizations through the dimensions of social capital. Journal of
wards responsible AI.https://arxiv.org/abs/2008.07326. Accessed 1 Information Science,30(5), 448–458. https://doi.org/10.1177/
Nov 2020 0165551504046997
Shaban-Nejad, A., Michalowski, M., & Buckeridge, D. L. (2018). Health Wu, L., Chuang, C. H., & Hsu, C. H. (2014). Information sharing and
intelligence: how artificial intelligence transforms population and collaborative behaviors in enabling supply chain performance: A
personalized health. NPJ Digital Medicine,1, 53. https://doi.org/10. social exchange perspective. International Journal of Production
1038/s41746-018-0058-9 Economics,148, 122–132. https://doi.org/10.1016/j.ijpe.2013.09.
Singer, P. W. (2020). Can AI systems be ethical? In Yorke-Smith, N. 016
(Eds.), Responsible artificial intelligence: how to develop and use AI Ye, Y. (2010). Correlates of consumer trust in online health information:
in a responsible way, 78–99 findings from the health information national trends survey. Journal
Singh, P., Dwivedi, Y. K., Kahlon, K. S., Sawhney, R. S., Alalwan, A. A., of Health Communication,16(1), 34–49. https://doi.org/10.1080/
& Rana, N. P. (2020). Smart monitoring and controlling of govern- 10810730.2010.529491
ment policies using social media and cloud computing. Information Zaheer, N., & Trkman, P. (2017). An information sharing theory perspec-
Systems Frontiers,22(2), 315–337 tive on willingness to share information in supply chains. The
Stieglitz, S., Meske, C., Ross, B., & Mirbabaie, M. (2020). Going back in International Journal of Logistics Management,28(2), 417–443.
time to predict the future-the complex role of the data collection https://doi.org/10.1108/IJLM-09-2015-0158
period in social media analytics. Information Systems
Frontiers,22(2), 395–409
Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
Sumner, S. A., Galik, S., Mathieu, J., Ward, M., Kiley, T., Bartholow, B.,
tional claims in published maps and institutional affiliations.
& Mork, P. (2019). Temporal and geographic patterns of social