It’s complicated: People emotionally tied to robots can undermine relationships with co-workers

Subgroup formation in human–robot teams: A multi-study mixed-method approach with implications for theory and practice. Sangseok You,Lionel P. Robert. Journal of the Association for Information Science and Technology, February 10 2022. https://doi.org/10.1002/asi.24626

Abstract: Human–robot teams represent a challenging work application of artificial intelligence (AI). Building strong emotional bonds with robots is one solution to promoting teamwork in such teams, but does this come at a cost in the form of subgroups? Subgroups—smaller divisions within teams—in all human teams can undermine teamwork. Despite the importance of this question, it has received little attention. We employed a mixed-methods approach by conducting a lab experiment and a qualitative online survey. We (a) examined the formation and impact of subgroups in human–robot teams and (b) obtained insights from workers currently adapting to robots in the workplace on mitigating impacts of subgroups. The experimental study (Study 1) with 44 human–robot teams found that robot identification (RID) and team identification (TID) are associated with increases and decreases in the likelihood of a subgroup formation, respectively. RID and TID moderated the impacts of subgroups on teamwork quality and subsequent performance in human–robot teams. Study 2 was a qualitative study with 112 managers and employees who worked collaboratively with robots. We derived practical insights from this study that help situate and translate what was learned in Study 1 into actual work practices.

6 DISCUSSION

Work teams relying on AI-enabled technology such as robots will increase along with efforts to design them to elicit strong emotional bonds from humans (Vreede & Briggs, 2019). Our results underline the need for a more cautious and considerate approach when attempting to elicit such responses. Next, we discuss implications for theory and work.

6.1 Theoretical implications

Our findings have several implications for theory and research. First, Study 1 highlighted the overall need to extend theories of subgroup formation to include AI-enabled technologies. Study 1 found evidence of subgroup formation in human–robot teams and its negative impacts. Previous research demonstrated that strong bonds with technology could benefit individuals (K.-K. Kim et al., 2010; Read et al., 2011). Our work goes beyond this by demonstrating whether and when such strong bonds can be problematic. In doing so, we highlight the need for greater theoretical attention. There is a vast body of research on subgroup formation, and it is not clear how much of it is directly applicable to human–robot teams.

Second, Study 1's findings contribute to the existing theories on subgroups. We found that subgroups positively impact teamwork quality when robot identification is low, but these subgroups become problematic when robot identification is high. However, subgroup formation led to increased teamwork quality when TID was high. Our findings identified the particular mechanisms that can help dictate when subgroups are likely to increase/decrease teamwork quality. These findings answer recent calls to identify moderators that constrain the impacts of subgroups (Meyer et al., 2015; Thatcher & Patel, 2012).

6.2 Work practice implications

First, it is clear that the inclusion of robots can lead to subgroups. The surveyed workers had either already seen evidence of subgroups in human–robot work collaborations or were genuinely afraid they would see them. To address these concerns, many of the workers suggested ways to promote communication, training, and team-building that could all be labeled as approaches to promote TID. This seems to align with our findings from Study 1 regarding the importance of TID in reducing the negative impacts of subgroups. Study 2 goes beyond Study 1 by situating and translating the TID construct to changes in actual work practices.

Second, it was interesting that workers' comments were primarily directed at promoting human relationships rather than reducing human-to-robot bonds. It is possible that the workers' suggestions relating to the promotion of TID could be viewed as ways to reduce RID, but this was not directly suggested by workers. Instead, workers might actually see the value of strong RID and not want to undermine it but instead were more concerned with promoting TID through stronger human bonds. If so, this finding aligns well with Study 1's finding that RID can actually have a positive impact as long as it is coupled with strong TID.

Third, there was an emphasis on leadership in addressing issues associated with subgroups. On the one hand, workers thought that strong leadership was needed to help workers understand the value of robots while promoting strong bonds among human coworkers. On the other hand, managers indicated that strong leadership was needed to determine who should be assigned what tasks. That being said, the importance of leadership in human–robot teams is a relatively unexplored area. Despite this, Study 2's findings seem to imply that human–robot team scholars need to broaden their research agenda to include leadership.

Finally, it seems that as AI and the world of work are reshaping the meaning of work, employees would rather not lose what it means to be a human. More specifically, workers indicated the need to emphasize human contributions as distinct and separate from robot contributions. They also thought it was important to draw clear lines between robots and humans and to remind their coworkers of these distinctions. In doing so, workers latently suggested that organizations must highlight the uniqueness of being a human and instill this value through training, communication, and evaluation. To this point, our results have broader implications for our understanding of how AI reshapes the world of work.

6.3 Limitations and future research

The study has several limitations. This study employed one method to measure subgroup formation among other possible approaches. We used cohesion between humans and robots to measure subgroups because this aligns with the paper's definition of subgroups. However, previous research has employed many different subgroup measures (Polzer et al., 2006; Thatcher & Patel, 2012). Future research could investigate different measures. Second, while the experimental study found evidence of subgroups, the brief interaction is a limitation. Future research could examine how emotional bonds evolve over time (Björling et al., 2020). Third, robots used in Study 1 were not fully autonomous, representing many of the robots used in workplaces today (i.e., cobots). Future studies could vary the robot's autonomy to determine its importance in understanding subgroup formation and its impacts. Finally, our experimental study was designed to ensure internal validity at the expense of external validity. Future studies should examine the phenomena in field settings to qualitatively unpack the bonding process over time.