=Paper=
{{Paper
|id=Vol-3276/SSS-22_FinalPaper_87
|storemode=property
|title=AI Agents for Facilitating Social Interactions and
Wellbeing
|pdfUrl=https://ceur-ws.org/Vol-3276/SSS-22_FinalPaper_87.pdf
|volume=Vol-3276
|authors=Hiroaki Hamada,Ryota Kanai
}}
==AI Agents for Facilitating Social Interactions and
Wellbeing ==
https://ceur-ws.org/Vol-3276/SSS-22_FinalPaper_87.pdf
AI Agents for Facilitating Social Interactions
and Wellbeing
Hiro Taiyo Hamada1, Ryota Kanai2
1,2
Araya Inc.
hamada_h@araya.org, kanair@araya.org
Abstract vision, and natural language processing (NLP). Multiple
Wellbeing AI has been becoming a new trend in individ- services have been introduced to analyze or intervene in
uals' mental health, organizational health, and flourishing mental health by accessing peoples' emotions. For example,
our societies. Various applications of wellbeing AI have analyses by personal genetic data (Fang et al. 2020), images
been introduced to our daily lives. While social relation- (Reece et al. 2017), and texts in social media (Chancellor et
ships within groups are a critical factor for wellbeing, the
development of wellbeing AI for social interactions re- al. 2020) predict risks and mental conditions including men-
mains relatively scarce. In this paper, we provide an over- tal disorders. Some applications further intervene in mental
view of the mediative role of AI-augmented agents for so- conditions based on theories of psychological intervention
cial interactions. First, we discuss the two-dimensional such as cognitive-behavioral therapies (CBTs; van Agteren
framework for classifying wellbeing AI: individual/group et al. 2021). Although social factors are known to be crucial,
and analysis/intervention. Furthermore, wellbeing AI
touches on intervening social relationships between hu- most AI applications for wellbeing focus on individuals and
man-human interactions since positive social relation- much less on social groups. Given that we spend most of our
ships are key to human wellbeing. This intervention may time in multiple social groups such as family, workplaces,
raise technical and ethical challenges. We discuss oppor- schools, social clubs, etc., the opportunities and potential
tunities and challenges of the relational approach with impact of such group-targeted AI applications would be
wellbeing AI to promote wellbeing in our societies.
enormous. However, AI applications of social groups for
wellbeing have attracted little attention.
Introduction Here, we present an overview of the emergent role of AI-
augmented agents for social interactions. First, we investi-
COVID-19 has revealed the importance of the sense of be- gate the literature on psychological wellbeing and provide a
longingness and loneliness in mental health of our societies two-dimensional classification of AI-augmented agents: in-
(COVID-19 Mental Disorders Collaborators, 2021). Well- dividual/group and analysis/intervention. The first dimen-
being has attracted attention of psychology and public health sion concerns whether wellbeing AI is used for the analysis
for improving the mental health of individuals and organi- or the intervention. The second dimension focuses on
zations and has become one of the main targets for public whether an AI-augmented agent targets individuals or
health organizations such as the World Health Organization groups. We point out opportunities for the recently emerg-
(WHO; Topp et al. 2015). ing approach, the so-called relational approach, where AI-
Wellbeing has been studied intensively in the context of augmented agents are applied to human-human interactions
psychology (Andrews et al. 1976; Diener et al. 2018; Topp within groups. Finally, we discuss challenges in the rela-
et al. 2015). In psychology, multiple constructs of wellbeing tional approach of AI- augmented agents. We shed light on
have been proposed (Dodge et al. 2012). For example, Ryff broader opportunities for AI-augmented agents, and high-
and Keyes proposed that wellbeing is composed of multiple light technological and ethical challenges for promoting
factors such as autonomy, environmental mastery, personal wellbeing in the real and virtual societies.
growth, positive relations with others, purpose in life, and
self-acceptance (Ryff and Keyes 1995). Although there are
differences in emphasis among psychological theories, pos- Social Construct of Wellbeing
itive social relationships have been identified as a crucial
factor. The notion of wellbeing has attracted attention in the context
There is a new trend to apply artificial intelligence (AI) to of healthy individual lives and societies. Subjective wellbe-
enhance wellbeing due to the development of emotion anal- ing (SWB) has been widely measured as a screening tool for
ysis technologies such as genome-wide analysis, computer mental disorders based on self- reported questionnaires such
___________________________________
In T. Kido, K. Takadama (Eds.), Proceedings of the AAAI 2022 Spring Symposium
“How Fair is Fair? Achieving Wellbeing AI”, Stanford University, Palo Alto, California,
USA, March 21–23, 2022. Copyright © 2022 for this paper by its authors. Use permitted
under Creative Commons License Attribution 4.0 International (CC BY 4.0).
31
as the WHO questionnaire. Several models of SBW have that interventions including exercises and psychological in-
been proposed (Andrews et al. 1976, Dodge et al. 2012, tervention for social relationships on wellbeing can promote
Topp et al. 2015). SWB is composed of multiple facets com- wellbeing and prevent mental illness.
prising two components (Schimmack et al. 2008; Luhmann We spend many hours with family, friends, and col-
2012): affective and cognitive evaluations of one's life. The leagues. Subjective wellbeing in groups such as working
affective evaluation measures the emotional experiences of place is also studied well (Harter et al. 2003; Jain et al. 2009).
people in daily lives while the cognitive evaluations meas- Working environments and social networks influence well-
ure how people evaluate their lives based on their ideals. The being and healthy behaviors. Associations between work en-
affective and cognitive aspects are associated with different vironment and wellbeing are known (Harter et al. 2003;
scales such as daily emotional experience and life satisfac- Bowling et al. 2010). Life satisfaction and other related fac-
tion, respectively (Diener et al. 2018). Recent studies also tors such as job satisfaction and positive affect are related to
suggested that another supplementary factor, harmony in wellbeing. Another evidence further showed that mindful-
life in a social context is also associated with SWO (Kjell et ness training had small-to-moderate effects on psychologi-
al. 2016). Harmony in life reflects social and environmental cal distress, wellbeing, and sleep (Bartlett et al. 2019) alt-
situations and is associated with psychological balance and hough the influence on work performance could not be con-
flexibility in life. Therefore, social factors play a critical role cluded due to the insufficiency of pooled data. Internet-
in SWO. based interventions on workers showed small-to-moderate
Social personalities for wellbeing have been widely stud- effects on work effectiveness and psychological wellbeing
ied, showing consistent results. A recent meta- analysis, for in workplaces (Carolan et al. 2017).
example, revealed that widely used personality factors (e.g. Psychological interventions on social networks, so- called
NEO-PI-R and HEXACO questionnaires) are correlated social network interventions, are also effective on wellbeing
with several aspects of SWO such as life satisfaction, posi- (Hunter et al. 2019). This relatively new approach cares for
tive/negative affect, and positive relation with others (An- changes in information flow by intensifying, deleting, and
glim et al.. 2020). The study especially found that these as- transferring social ties (Valente 2012). The social network
pects of SWO are positively correlated with extraversion intervention is expected to enhance the effectiveness of
and conscientiousness although negatively correlated with health outcomes such as lower drug use, healthy sex behav-
neuroticism. The sensitivity of SWO, thus, could reflect the iors, stronger social support, and wellbeing.
personality traits of subjects. It is noteworthy that extraver- It is interesting to ask whether this approach is useful for
sion, as well as neuroticism and conscientiousness, also in- social media and online gameplay. There is a strong public
fluence related factors like depressive symptoms (Hakulinen interest in the association of social media use and game
et al. 2015). Extraversion is a social indicator for higher pos- playing with mental health. Their potentially harmful influ-
itive relationships with others. Meanwhile, neuroticism is a ences on mental health have often drawn public attention
social indicator for less positive relationships related to (Huang 2010; Prescott et al. 2018), but the relationship re-
loneliness (Buecker et al. 2020). The association between mains unclear, perhaps due to huge differences in design and
these personalities and wellbeing- related factors supports concepts within social media (Sakurai et al. 202) and games
the idea to promote wellbeing via positive relationships. (Johannes et al. 2021). Communication within online video
It is an important question whether behavioral practice games such as e- sports can be essential for effective team
can change social relationships and wellbeing. Multiple at- performance. Effective social intervention may increase not
tempts showed enhancement of wellbeing as well as associ- only team performance and wellbeing, but the potential of
ated factors by healthy behaviors such as exercising (Che- such social interventions remains clear. Findings on group
kroud et al. 2018) and psychological interventions (van Ag- wellbeing, nonetheless, reveal another potential target of
teren et al. 2021). A cross-sectional study from 1.2 million interventions in our societies.
individuals in the U.S. showed that physical exercising rou- To sum up, existing literature revealed associations of so-
tines such as popular team sports, aerobic, and gym activi- cial factors related to genetics, environments, and behaviors
ties decrease up to 22% of mental health burdens compared with wellbeing. These multiple findings clarify possibilities
to the non-exercising group (Chekroud et al. 2018). of interventions of subjective wellbeing as well as group
Furthermore, different psychological interventions such wellbeing.
as behavioral activation interventions (BA), positive psy-
chological intervention (PPI), and mindfulness-based inter-
ventions (MBI) also showed small-to-moderate effects on Types of Wellbeing AI
wellbeing (van Agteren et al. 2021). Social interventions al- The effectiveness of interventions on wellbeing triggered
leviated social isolation (Dickens et al. 2011) and loneliness expectations to conduct research and development along
(Masi et al. 2010). These empirical findings further support with a trend of digital therapeutics. Digital therapeutics are
32
Individ- Analysis/Intervention Categories Examples
ual/Group
Individual Analysis Genetics Ø Depression Risk (Fang et al. 2020)
Ø Subjective Wellbeing (Røysamb et al. 2018)
Ø Social Support (Wang et. al. 2017)
Individual Analysis Mental Health Ø Emotion Detection (Canedo et al. 2019)
Ø Screening Mental Health Status on Social Me-
dia (Chancellor et al. 2020)
Individual Intervention Health Care Ø Behavioral Cognitive Therapy (Woebot,
(Ahmed et al. 2021; Du- Todaki; Jang et al. 2021)
radoni et al. 2021) Ø Promotion for Mental Wellbeing (Shim; Ly et
al. 2017)
Ø Cancer Cares for Young Survivors (Vivibot;
Greer et al. 2019)
Individual Intervention Workplace Ø A Chatbot for Improvement for Sedentary Be-
haviour and Wellbeing (Welbot; Haile et al.
2020)
Group Analysis Emotion Analysis Ø Images (Tan et al. 2017)
(Veltmeijer et al. 2020) Ø Sounds (Franzoni et al. 2020)
Ø Videos (Sánchez et al. 2020)
Ø Social Media (Gong et al. 2019)
Group Intervention Mental Health Ø A Chatbot for Positive Messaging (Sunny; Na-
rain et al. 2020)
Group Intervention Group Discussion Facili- Ø Group Discussion Facilitation such as
tation GlahBlahBot(Shin et al.. 2020), Micbot(Ten-
nent et al.. 2019), Groupfeedbot (Kim et al.
2020), Vulnerable-Robots (Traeger et al.
2020)
Table 1: A list of examples for analysis and interventions for wellbeing-related factors in different categories.
evidence-based therapeutic interventions with software pro- groups (Narain et al. 2020). We belong to multiple social
grams to cure and prevent medical disorders. There is also a groups in different contexts such as schools, workplaces,
trend to apply AIs to such evidence-based interventions in sports clubs, and our houses. Scopes of group wellbeing
mental health (D’Alfonso 2020). In this section, we summa- should be also broad. There are some studies on group well-
rize and explain the types of AI-augmented applications being and proposals to promote such group wellbeing from
such as robots, avatars, and bots for wellbeing. By doing so, AI-augmented agents such as robots (Kim et al. 2020; Shin
we clarify currently active approaches in wellbeing AI. et al. 2020; Tennent et al. 2019; Traeger et al. 2020). How-
We categorize AI applications for wellbeing with two-di- ever, group wellbeing targeted by AI applications is rela-
mensional axes: analysis/intervention and individual/group tively understudied and may bring new opportunities for the
(Table 1.). The first dimension means whether the aim of promotion of wellbeing.
digital wellbeing is analysis or intervention. Many potential In summary, most of the applications in wellbeing AI fo-
applications focus on monitoring emotional states or related cus on analysis and psychological interventions of individ-
factors associated with wellbeing through genes (Fang et al. ual wellbeing through mobile devices. Meanwhile, group
2020), images (Canedo et al. 2019), and texts (Chancellor et wellbeing is relatively under-explored but may have a huge
al. 2020) from social media. impact on our societies.
Other applications target the individual status of wellbe-
ing-related factors through apps. Meanwhile, positive social
networks are a crucial social basis for groups as well as in- The Relational Approach for Group Wellbe-
dividuals. There are only a few applications of wellbeing ing with AI
targeting groups from this perspective (Narain et al. 2020).
In this section, we overview a relational approach for group
For example, a Facebook messenger chatbot, Sunny, is
wellbeing with the literature of analysis and interventions on
meant to promote social interactions and wellbeing within
33
human-human interactions with AI agents. By doing so, we 2020). However, much fewer studies target social groups for
outline opportunities of wellbeing AI for group wellbeing. wellbeing. Some recent studies worked on discussion facil-
We then discuss types of the relational approach: analysis of itation with artificial agents (Kim et al. 2020; Traeger et al.
group dynamics and social connectedness. Finally, the chal- 2020; Tennent et al. 2019) and wellbeing promotion (Narain
lenges of the relational approach will be discussed. et al. 2020). The group intervened in a social group to induce
conversations and engagement on problem-solving. One
Literature Review on Analysis and Interventions study with a messenger chatbot, Sunny, worked on group
for Social Groups with AI wellbeing by sending positive messages to 4 member groups
and had positive effects on wellbeing (Narain et al. 2020).
Detection and intervention of group wellbeing with AI are
These studies are mostly limited in discussion facilitation,
not well studied. However, related studies on automated
but potential applications of social groups can be more ex-
group emotion and artificial agents for social interactions
tensive in different fields such as houses, schools, hospitals,
have been active recently.
caregivers, sports, workplaces, social media, tourism, where
Automated group-level emotion recognition has been
social groups are formed. For example, artificial agents
studied recently since 2012 (Veltmeijer et al. 2020; Table.
could work for team engagements in sports by giving anal-
1). Group emotion is not a simple sum of individual emo-
ysis or feedback based on their performance. The AI-aug-
tions in a group. Instead, automated emotion recognition
mented agents could also work on the mediation of conflicts
needs to track unique group emotion dynamics. A user sur-
between members in workplaces as well as enhancement of
vey has been developed as a proxy of such group emotion
discussion facilitation. In doing so, we may expect artificial
(Dhall et al. 2015). Multiple studies predict group emotional
agents to promote wellbeing.
labels based on various datasets from images (Tan et al.
It is also critical to ask whether we explicitly need such
2017), videos (Sánchez et al. 2019), and social media (Gong
artificial agents. It may be sufficient to have AIs without
et al. 2019). Such studies target different sizes and states of
agents such as recommendation systems for e-commerce.
seated, standing, and dynamic groups. Veltmeijer et al.
One benefit of artificial agents could be related to attentional
pointed out three technical challenges (Veltmeijer et al.
engagements by agents (Chae et al. 2016; Lucas et al. 2014;
2020). First, group size changes. Second, subgroup emo-
Mollahosseini et al. 2018; Spicer et al. 2021). Multiple stud-
tions in a larger group can be different. Third, group emotion
ies showed artificial agents enhance engagements (Oertel et
can also change. Although methods are under development,
al. 2020). This attentional engagement can be augmented by
automated emotion detections for groups are perhaps appli-
the appearance of artificial agents (Li et al. 2010; Bente et
cable to group wellbeing detection. Several types of re-
al. 2008). Several pieces of evidence also revealed that the
searches, applications, and commercial products for interac-
appearance of artificial agents influences human trust of the
tions with AI have been introduced in various situations
agents and induces similar human behaviors to humans by
such as education, hospitals, games, workplaces, social me-
the agents (Caruana et al. 2017; Lucas et al. 2014). It is un-
dia, banks, online dating, sports, tourism, etc. These agents
certain whether artificial agents work best in all situations,
are expected to increase learning speed, team performance,
but they may exert stronger influences than just non-agen-
successful dating matches, or satisfaction during traveling.
tive AIs via emulating human-like interactions.
Not only AI without agents but also AI-augmented agents
In sum, previous studies on the analysis of emotional de-
are widely used in our societies. We define such artificial
tection for social groups and intervention of social interac-
agents as three types: robots, social bots, and avatars. AI-
tions are active. However, these analyses and interventions
augmented agents are commonly used for cooperative pur-
on social interactions have not yet merged, and few studies
poses for interactions between human and artificial agents.
focus on group wellbeing.
Human-robot studies are commonly done to understand the
capability of robots (Sheridan, 2016) and how humans rec-
ognize robots (Chae et al. 2016; Lucas et al. 2014). Social Communicating with Social Human Groups via
bots have also been studied for communications through Artificial Agents
apps and social media (Assenmacher et al. 2020). Avatar- The mediative role for human-human interactions with arti-
human interactions are further studied in the context of re- ficial agents has not been well studied. Potential opportuni-
mote learning although humans control such avatars in most ties of such artificial agents are more extensive than current
studies (Chae et al. 2016). These studies aim for interactions opportunities for individual wellbeing. However, the medi-
between human and artificial agents. ative role of AI in group wellbeing, the so-called relational
An emergent application of artificial agents as social me- approach, has not been explicitly explored. Here, we clarify
diators is expected to promote social interactions between two types of a relational approach to social groups. By doing
humans and prevent problematic behaviors within a group so, we prompt the development of relational approaches for
(Chita-Tegmark, 2020; Dafoe et al. 2021; Rahwan et al. group wellbeing.
34
One type of the relational approach is to analyze group and successful introduction and management of wellbeing
dynamics itself from conversations or their behaviors (Fig- AI.
ure.1.A). The previous studies on automated emotion detec- Fairness of computation is raised as an important issue in
tion target analyzing such group dynamics by facial expres- AI research. Fairness in AI research is composed of three
sions and conversations (Kim et al. 2020; Narain et al. 2020, perspectives: fairness, conflict of interest, and respect of dif-
Tennent et al. 2019; Traeger et al. 2020). A robot agent ferent communities. First, each social connection within a
study also targets group performance such as total conver- group should be fairly considered. Asymmetrical social con-
sation time (vulnerable-robots; Traeger et al. 2020). This ap- nections may cause issues within a group. Next, the intro-
proach focuses on average or wholistic group dynamics not duction of wellbeing AI by administrators should be fairly
considering the relationship among members in a group. considered for users. Wellbeing AI can be expected by ad-
Another type of the relational approach is to analyze one- ministrators to enhance engagements of users in workplaces,
to-one member interactions in detail (Figure.1.B). We also social media and games. Such increased engagements have
directly contact a person in management not by groups. For the potential to deteriorate life satisfaction causing burnout
example, when certain group members engage in group dis- symptoms in the long term. Long-term wellbeing for users,
cussion, another member familiar with one of the members then, should be considered. Third, different cultures of com-
may have insight. In this case, you as a mediator want to ask munities should be respected. Perception of wellbeing is
the member to promote his or her engagement in the discus- known to differ in different communities and populations
sion. Such a role can be served by an artificial agent. This (Lai et al. 2013). This example may reflect individual traits
type should be computationally intensive since N-to-N hu- based on experience, personality, and genetics. Computa-
man interactions are analyzed based on methods such as tion of social connections should not only consider a specific
computer vision and natural language processing (Poria et type of individual traits but also multiple perceptions.
al. 2019). Along with such development, computation Privacy of human-human interactions is another crucial
power increases these days, so current computation power issue. Multiple issues have been raised by previous actions
could be sufficient for human- human interactions within a by companies on controlling and using the private data of
few members.
It is interesting to ask whether these two types of the re-
lational approaches can be integrated as computational
methods like social network analysis (Gesell et al. 2013) and
network controllability (Liu et al. 2011). One potential key
field is related to network analysis considering both each so-
cial connection and network organization. In neuroscience,
analysis and intervention of whole-brain state with region-
region interactions are actively studied (Tang and Bassett
2018). Such network analysis further would bring an inte-
grative perspective of social interactions and group dynam-
ics. These approaches should be enriched by the develop-
ment of the two types of the relational approach for group
wellbeing. The distinction could be tentative but should be
useful to work on wellbeing AI from a view of social inter-
actions.
Challenges on Relational Approach and Potential
Ethical Issues
We discuss three types of challenges on designing and man-
aging wellbeing AI for the human-human social interactions
based on the relational approach. 1) Changes related to fair-
ness issues of computation and authority from the viewpoint
of different cultural contexts, conflict of interest, and struc- Figure 1: Intervention on group dynamics. A. AI agents
ture of benefits. 2) Challenges related to the privacy of hu- analyze group dynamics, and intervene a social group by
man-human interactions from the viewpoint of ownership communicating to all members. B. AI agents analyze one-
and autonomy of communications. 3) Challenges related to to-one interactions within a social group and intervene so-
usefulness from the viewpoint of users of accessibility and cial connections or specific members based on member’s
safety. These challenges must be overcome for the effective connections.
35
users. In this regard, privacy, autonomy, and ownership of interactions by analyzing and managing human-human in-
social interactions should be considered. First, excessive ac- teractions for the introduction of AI-supported wellbeing in
cess and storage of private data should not be permissible. the era of digital worlds. First, we described psychological
Communications are performed with image, auditory, and research on wellbeing based on personality, genetics, and
text information but storing, analyzing, and providing to a behavioral and cognitive interventions, and concluded that
third party should depend on the permission of users. Such social relationships are crucial for wellbeing. We, then,
actions potentially causing disadvantages to users should identified an unexplored category of wellbeing AI and group
not further be taken. Second, whether actions by wellbeing wellbeing. Group wellbeing through telecommunications is
AI are excessive should be considered. Such actions may especially critical since the expansion of telecommunica-
cause behavioral constraints for users. Such interventions to tions may cause psychological issues such as distress and
corruption of autonomy in groups should not be permissible. loneliness which are reported during COVID-19.
Third, appropriate interventions on social connections By reviewing previous literature of interventions on so-
should be considered. Related to autonomy preservation, cial networks with a robot and virtual agents, we further in-
some interventions may be permissible depending on group troduced the relational approach, which analyzes and medi-
characteristics, but others not. This is perhaps related to a ates human-human interactions with artificial agents such as
discussion of moral agency in AI where what AI agents are chatbots and robot agents. The relational approach mediates
allowed to perform. Members within groups should deter- human-human social interactions in the real or digital world
mine which type of analysis and interventions is permissible. to promote wellbeing and other factors such as team perfor-
Whether interventions to users are appropriate or not is mance. Finally, we discussed potential challenges of design
critical. One reason is designing wellbeing environments and usage of the relational approach in wellbeing AI to es-
can be more important than introducing wellbeing AI. For tablish its successful support of human social networks.
example, the relationship between employees and wellbeing We shed light on the mediative roles of AI-augmented
is dependent on working environments. In other words, it agents to benefit human mental health and wellbeing in real
might not be important to have such wellbeing AI if the and digital environments. By doing so, we expect a broader
working environment is not designed to promote wellbeing. understanding and further development of group wellbeing.
This idea might be aligned with wellbeing in the importance
of game design rather than the importance of introducing
wellbeing AI. Appropriateness of wellbeing should be con- Acknowledgments
sidered from usability, understanding, and the public inter- We acknowledge Toshifumi Sasaki in Osaka University for
est of users. The stability of wellbeing AI is the priority to discussion.
be considered. Attempts to have wellbeing AI is still under
exploration. Real applications might not often be welcomed
in social contexts. What factors determine such usability for References
users should be investigated. Second, understanding users is Andrews, F.M.; Withey, S.B. 1976. Measuring Global Wellbeing.
important. Related to usability, the mismatch between users In Social Indicators of Wellbeing. New York: Plenum Press.
and applications might be associated with misunderstanding Anglim, J.; Horwood, S.; and Smilie, L.D. 2020. Predicting Psy-
of users by administrators. Third, a perspective of public in- chological and Subjective Wellbeing from Personality: A Meta-
terest is needed. This is a third-party view of wellbeing AI. Analysis. Psychological Bulletin 146(4): 279–323.
Even though users and administrators gain benefits from ap- doi.org/10.1007/978-1-4684-2253-5_3.
plications of wellbeing AI, the relational approach may have Assenmacher, D. et al.. 2020. Demystifying Social Bots: On the
a huge harmful impact on the public interest. Such appropri- Intelligence of Automated Social Media Actors. Social Media +
Society. doi.org/10.1177/2056305120939264
ateness should be considered too.
Multiple challenges including three types of perspectives Bente, G. et al. 2008. Avatar- Mediated Networking: Increasing
Social Presence and Interpersonal Trust in Net-Based Collabora-
exist for designing and managing the relational approach of tions. Human Communication Research 34(2): 287–318.
wellbeing AI since such approach is implicitly under devel- doi.org/10.1111/j.1468-2958.2008.00322.x
opment. Nonetheless, the relational approach of wellbeing Bartlett, L. et al. 2019. A Systematic Review and Meta-Analysis of
AI has huge room to benefit our societies. Workplace Mindfulness Training Randomized Controlled Trials.
Journal of Occupational Jealth Psychology 24(1): 108–126.
doi.org/10.1037/ocp0000146
Conclusion Bowling, N.A.; Eschleman, K.J.; and Wang, Q. 2011. A Meta-An-
alytic Examination of the Relationship Between Job Satisfaction
In this paper, we introduced the notion of AI-supported and Subjective Well-being. Journal of Occupational and Organiza-
wellbeing in the era of digital worlds and presented an over- tional Psychology 83(4): 915-34.
view of the relational approach to promoting positive social doi.org/10.1348/096317909X478557
36
Buecker, S. et al. 2020. Loneliness and the Big Five Personality Fang, Y. et al. 2020. Genomic Prediction of Depression Risk and
Traits: A Meta-analysis. European Journal of Personality 34(1): 8- Resilience under Stress. Nature Humam Behaviour 4: 111–8.
28. doi.org/10.1002/per.2229 doi.org/10.1038/s41562-019-0759-3
Chae, S.W.; Lee, K.C.; Seo, and Y.W. 2016. Exploring the Effect Franzoni, V.; Biondi, G.; and Milani, A. 2020. Emotional Sounds
of Avatar Trust on Learners’ Perceived Participation Intentions in of Crowds: Spectrogram-Based Analysis Using Deep Learning.
an e-Learning Environment. International Journal of Human-Com- Multimedia Tools and Applications 79: 36063–75.
puter Interaction 32(5): 373-93. doi.org/10.1007/s11042-020-09428-x
doi.org/10.1080/10447318.2016.1150643 Gesell, S.B.; Barkin, S.L.; and Valente, T.W. Social Network Di-
Canedo, D.; and Neves, A.J.R. 2019. Facial Expression Recogni- agnostics: a Tool for Monitoring Group Interventions. Implement
tion Using Computer Vision: A Systematic Review. Applied Sci- Science 8: 116. doi.org/10.1186/1748-5908-8-116
ences 9(21): 4678. doi.org/10.3390/app9214678 Gong, V.X. et. el. 2019. Estimate Sentiment of Crowds from Social
Carolan, S.; Harris, P.R.; and Cavanagh, K. 2017. Improving Em- Media during City Events. Transportation Research Record
ployee Well-Being and Effectiveness: Systematic Review and 2673(11):836-50. doi.org/10.1177/0361198119846461
Meta-Analysis of Web-Based Psychological Interventions Deliv- Greer, S. et al. 2019. Use of the Chatbot “Vivibot” to Deliver Pos-
ered in the Workplace. Journal of medical Internet research 24(1): itive Psychology Skills and Promote Well-Being Among Young
108–26. doi.org/10.1037/ocp0000146 People After Cancer Treatment: Randomized Controlled Feasibil-
Caruana, N.; Spirou, D.; and Brock, J. Human Agency Beliefs In- ity Trial. JMIR Mhealth Uhealth 7(10): doi.org/10.2196/15018
fluence Behaviour During Virtual Social Interactions. PeerJ Haile, C. et al. 2020. Pilot Testing of a Nudge-Based Digital Inter-
5:e3819. doi.org/10.7717/peerj.3819 vention (Welbot) to Improve Sedentary Behaviour and Wellbeing
Chancellor, S.; and De Choudhury, M. 2020. Methods in Predictive in the Workplace. International journal of environmental research
techniques for Mental Health Status on Social Media: a Critical and public health 17(16): 5763. doi.org/10.3390/ijerph17165763
review. npj Digital Medicine 3(43). doi.org/10.1038/s41746-020- Harter, J.K.; Schmidt, F.L.; and Keyes, C.L.M. 2003. Well-being
0233-7 in the Workplace and its Relationship to Business Putcomes: A Re-
Chekroud, S.R. et al. 2018. Association between Physical Exercise view of the Gallup Studies. Flourishing: Positive psychology and
and Mental Health in 1·2 Million Individuals in the USA between the life well-lived, edited by C.L.M. Keyes & J. Haidt, 205–24.
2011 and 2015: a Cross-Sectional Study. The lancet. Psychiatry Washington D.C.: American Psychological Association.
5(9), 739–46. doi.org/10.1016/S2215-0366(18)30227-X Hakulinen, C. et al. 2015. Personality and Depressive Symptoms:
Chita-Tegmark, M.; and Scheutz, M. 2021. Assistive Robots for Individual Participant Meta-Analysis of 10 Cohort Studies. De-
the Social Management of Health: A Framework for Robot Design pression and Anxiety 32(7): 461-70. doi.org/10.1002/da.22376
and Human–Robot Interaction Research. International Journal of Huang, C. 2010. Internet Use and Psychological Well-being: a
Social Robotics 13: 197-217. doi.org/10.1007/s12369-020-00634- Meta-Analysis. Cyberpsychology, behavior and social networking
z 13(3): 241-9. /doi.org/10.1089/cyber.2009.0217
COVID-19 Mental Disorders Collaborators. 2021. Global Preva- Hunter, R.F. et al. 2019. Social Network Interventions for Health
lence and Burden of Depressive and Anxiety Disorders in 204 Behaviours and Outcomes: A Systematic Review and Meta-analy-
Countries and Territories in 2020 due to the COVID-19 pandemic. sis. PLoS medicine 16(9): e1002890. doi.org/10.1371/jour-
The Lancet 398(10312): 1700-12. doi.org/10.1016/S0140- nal.pmed.1002890
6736(21)02143-7
Jain, A.K.; Giga, S.I.; and Cooper, C.L. 2009. Employee Wellbe-
Dafoe, A. et al. 2021. Cooperative AI: Machines Must Learn to ing, Control and Organizational Commitment. Leadership & Or-
Find Common Ground. Nature 593: 33-36. ganization Development Journal 30(3): 256-73.
doi.org/10.1038/d41586-021-01170-0 doi.org/10.1108/01437730910949535
D’Alfonso, S. 2020. AI in Mental Health. Current Opinion Psy- Jang, S. et al. 2021. Mobile App-based Chatbot to Deliver Cogni-
chology 36: 112-7. doi.org/10.1016/j.copsyc.2020.04.005 tive Behavioral Therapy and Psychoeducation for Adults with At-
Dhall, A. et al. 2015. The More the Merrier: Analysing the Affect tention Deficit: A Development and Feasibility/Usability Study.
of a Group of People in Images. 11th IEEE International Confer- International journal of medical informatics 150: 104440.
ence and Workshops on Automatic Face and Gesture Recognition doi.org/10.1016/j.ijmedinf.2021.104440
(FG). Johannes, N.; Vuorre, M.; and Przybylski, A.K. 2021. Video Game
Dickens, A.P. et al. 2011. Interventions Targeting Social Isolation Play is Positively Correlated with Well-being. Royal Society Open
in Older People: a Systematic Review. BMC Public Health 11(647). Science 8(2). doi.org/10.1098/rsos.202049
doi.org/10.1186/1471-2458-11-647 Kjell, O.N.E. et al. 2016. The Harmony in Life Scale Complements
Diener, E.; Oishi, S.; and Tay, L. 2018. Advances in Subjective the Satisfaction with Life Scale: Expanding the Conceptualization
Well-Being Research. Nature Human Behaviour 2, 253–60. of the Cognitive Component of Subjective Well-Being. Social In-
doi.org/10.1038/s41562-018-0307-6 dicators Research 126: 893-919. doi.org/10.1007/s11205-015-
Duradoni, M. et al. 2021. Robotic Psychology: A PRISMA Sys- 0903-z
tematic Review on Social-Robot-Based Interventions in Psycho- Kim, S. et al. 2020. Bot in the Bunch: Facilitating Group Chat Dis-
logical Domains. J 4(4):664-97. doi.org/10.3390/j4040048 cussion by Improving Efficiency and Participation with a Chatbot.
Dodge, R. et al. 2012. The Challenge of Defining Wellbeing. In- In Proceedings of the 2020 CHI Conference on Human Factors in
ternational Journal of Wellbeing 2(3): 222-35. Computing Systems. Honolulu, HI, April 25-30.
doi:10.5502/ijw.v2i3. Lai, L.C.H.; Cummins, R.A.; and Lau A.L.D. 2013. Cross-Cultural
Difference in Subjective Wellbeing: Cultural Response Bias as an
37
Explanation. Social Indicators Research 114: 607-19. on LINE, Facebook, Twitter, and Instagram. PLoS ONE 16(3):
doi.org/10.1007/s11205-012-0164-z e0246090. doi.org/10.1371/journal.pone.0246090
Li, D.; Rau, P.L.P.; and Li, Y. 2010. A Cross-cultural Study: Effect Sánchez, F.L. et al. 2020. Revisiting Crowd Behaviour Analysis
of Robot Appearance and Task. International Journal of Social Ro- through Deep Learning: Taxonomy, anomaly detection, crowd
botics 2: 175-86. doi.org/10.1007/s12369-010-0056-9 emotions, datasets, opportunities and prospects. Information Fu-
Liu, Y.Y.; Slotine, J.J.; and Barabási, A.L. 2011. Controllability of sion 64: 318-35. doi.org/10.1016/j.inffus.2020.07.008
Complex Networks 473: 167-73. doi.org/10.1038/nature10011 Schimmack, U. 2008. The Structure of Subjective Wellbeing. In
Lucas, G.M.; Gratch, J.; King, A.; and Morency, L. 2014. It’s only The science of subjective wellbeing, edited by M. Eid & R.J. Larsen,
a computer: Virtual humans increase willingness to disclose. Com- 97-123. New York: Guilford Press.
puters in Human Behavior 37: 94-100. Sheridan, T. 2016. Human-Robot Interaction: Status and Chal-
doi.org/10.1016/j.chb.2014.04.043 lenges. Human Factors 58(4): 525–32.
Luhmann, M. et al. 2012. Time Frames and the Distinction be- doi.org/10.1177/0018720816644364
tween Affective and Cognitive Well-being. Journal of research in Spicer, C.; Khwaounjoo, P.; and Cakmak, Y.Z. 2021. Human and
personality 46(4): 431-41. doi.org/10.1016/j.jrp.2012.04.004 Human-Interfaced AI Interactions: Modulation of Human Male
Ly, K.H.; Ly, A.M.; and Andersson, G. A Fully Automated Con- Autonomic Nervous System via Pupil Mimicry. Sensors (Basel
versational Agent for Promoting Mental Well-being: A Pilot RCT 21(4):1028. doi.og/10.3390/s21041028
Using Mixed Methods. Internet Interventions 10: 39-46. Shin, D. et al. 2021. BlahBlahBot: Facilitating Conversation be-
doi.org/10.1016/j.invent.2017.10.002 tween Strangers using a Chatbot with ML-infused Personalized
Masi, C.M. et al. 2011. A Meta-Analysis of Interventions to Re- Topic Suggestion. In Proceedings of the ACM SIGCHI Confer-
duce Loneliness. Personality and social psychology review: an of- ence on Human Factors in Computing Systems. New York: Asso-
ficial journal of the Society for Personality and Social Psychology ciation for Computing Machinery.
15(3) 219-66. doi.org/10.1177/1088868310377394 doi.org/10.1145/3411763.3451771
Mollahosseini, A. et al. Role of Embodiment and Presence in Hu- Tan, L. et al. 2017. Group Emotion Recognition with Individual
man Perception of Robots’ Facial Cues. International Journal of Facial Emotion CNNs and Global Image based CNNs. In Proceed-
Human-Computer Studies 116: 25-39. ings of the 19th ACM International Conference on Multimodal In-
doi.org/10.1016/j.ijhcs.2018.04.005 teraction. New York: Association for Computing Machinery.
doi.org/10.1145/3136755.3143008
Narain, J. et al. Promoting Wellbeing with Sunny, a Chatbot that
Facilitates Positive Messages within Social Groups. In Proceed- Tennent, H.; Shen, S.; and Jung M. 2019. Micbot: A Peripheral
ings of the ACM SIGCHI Conference on Human Factors in Com- Robotic Object to Shape Conversational Dynamics and Team Per-
puting Systems. New York: Association for Computing Machinery. formance. In Proceedings of ACM/IEEE International Conference
doi.org/10.1145/3334480.3383062 on Human-Robot Interaction (HRI). New York: Association for
Computing Machinery. doi.org/10.1109/HRI.2019.8673013
Oertel, C. et al. Engagement in Human-Agent Interaction: An
Overview. Frontiers in Robotics and AI 7: 92. Tang, E.; and Bassett, D.S. 2018. Colloquium: Control of Dynam-
doi.org/10.3389/frobt.2020.00092 ics in Brain Networks. Reviews of Modern Physics 90: 031003.
doi.org/10.1103/RevModPhys.90.031003
Poria, S. et al. 2018. MELD: A Multimodal Multi-Party Dataset for
Emotion Recognition in Conversations. In Proceedings of the 57th Topp, C.W. et al. 2015. The WHO-5 Well-Being Index: a System-
Annual Meeting of the Association for Computational Linguistics. atic Review of the Literature. Psychother Psychosom 84(3): 167-
Massachusetts: Association for Computational Linguistics. 76. doi.org/10.1159/000376585
doi.org/ 10.18653/v1/P19-1050 Traeger, M.L. et al. 2020. Vulnerable Robots Positively Shape Hu-
Prescott, A.T.; Sargent, J.D.; and Hull, J.G. 2018. Metaanalysis of man Conversational Dynamics in a Human–Robot Team. Proceed-
the Relationship between Violent Video Game Play and Physical ings of the National Academy of Sciences of the United States of
Aggression Over Time. Proceedings of the National Academy of America 117(12): 6370-5. doi.org/10.1073/pnas.1910402117
Sciences of the United States of America 115(40): 9882-8. Valente, T.W. 2012. Network Interventions. Science 337(6090):
doi.org/10.1073/pnas.1611617114 49-53. doi.org/10.1126/science.1217330
Rahwan, I.; Crandall, J.W.; and Bonnefon, J. 2020. Intelligent Ma- van Agteren, J. et al. 2021. A Systematic Review and Meta-Anal-
chines as Social Catalysts. Proceedings of the National Academy ysis of Psychological Interventions to Improve Mental Wellbeing.
of Sciences of the United States of America 117(14): 7555-7. Nature Human Behaviour 5: 631–52. doi.org/10.1038/s41562-
doi.org/10.1073/pnas.2002744117 021-01093-w
Reece, A.G.; and Danforth, C.M. 2017. Instagram Photos Reveal Veltmeijer, E.A.; Gerritsen, C; and Hindriks, K. 2021. Automatic
Predictive Markers of Depression. EPJ Data Science 6(15). Emotion Recognition for Groups: a Review. IEEE Transactions on
doi.org/10.1140/epjds/s13688-017-0110-z Affective Computing. doi.org/10.1109/TAFFC.2021.3065726
Røysamb, E.; and Nes, R.B. 2019. The Role of Genetics in Subjec- Wang, R.A.H. et al. 2017. Social Support and Mental Health in
tive Well-being. Nature Human Behaviour 3(3). Late Adolescence are Correlated for Genetic, as well as Environ-
doi.org/10.1038/s41562-018-0494-1 mental, Reasons. Scientific Reports 7: 13088.
Ryff, C.D.; and Keyes, C.L. 1995. The Structure of Psychological doi.org/10.1038/s41598-017-13449-2
Well-being Revisited. Journal of personality and social psychology
69(4): 719–27. doi.org/10.1037//0022-3514.69.4.719
Sakurai, R. et al. 2021. Who is Mentally Healthy? Mental health
profiles of Japanese social networking service users with a focus
38