Can AI Be Your Teammate or Friend? Frequent AI Users Are More Likely to Grant Humanlike Roles to AI

Abstract

The purpose of this research was to identify the comfort levels of professionals with AI in various humanlike roles. A survey of 787 full-time working adults showed that more active AI users are comfortable with AI in many humanlike roles, such as a teammate or a performance coach. Less active AI users, however, are uncomfortable with AI in these roles. Leaders, managers, and educators should prepare employees and students to responsibly address the social and psychological outcomes of increasingly humanlike AI.

Keywords

AI artificial intelligence AI as teammate Generative AI AI literacy emotional intelligence

Introduction

On October 16, 2023, at Gartner’s IT Symposium for CIOs, Gartner VPs Mary Mesaglio and Don Scheibenreif stated, “AI is the new machine and you’re in a relationship with it.” They suggested CIOs should help employees navigate their relationships with AI and suggested roles that people might grant AI included “consultant,” “protector,” “friend,” “coach,” “boss,” and “customer” (Mesaglio & Scheibenreif, 2023). Microsoft CEO Satya Nadella has on many occasions talked about his dream of all 8 billion people on earth having an AI tutor, an AI doctor, and an AI consultant (Singh, 2023). These public aspirations or predictions about granting humanlike roles to AI are increasingly common among business leaders. Scholars have considered the various roles professionals might grant to AI for many years (Andrews et al., 2023; Babic et al., 2020; Getchell et al., 2022; Seeber et al., 2020; Zhang et al., 2020), yet little is known about what most business professionals actually think about granting humanlike roles to AI. The purpose of this research is to explore emerging attitudes among professionals about comfort with granting various humanlike roles to AI.

Literature Review

We briefly review research about the roles that may be granted to AI. We first acknowledge research about anthropomorphism in AI and humanness ascribed to AI. Then, we ground most of the discussion in the work of two groups, Getchell et al. (2022) and Seeber et al. (2020), because of their focus on business communication and AI as teammates.

Until the launch of ChatGPT in 2022, most people’s encounters with AI chatbots were in consumer settings. Abundant literature exists about these consumer-facing chatbots. For many roles granted to AI, people may attribute humanness to the AI (Borau et al., 2021). Abundant research shows that anthropomorphizing bots—in name, appearance, gender, conversational ability, and other humanlike traits—generally increases humanness perceptions of AI (Borau et al., 2021; Hu et al., 2021; Park et al., 2023), although there are exceptions (Troshani et al., 2021). The current generation of AI contributes to perceptions of humanness due to enhanced conversational ability and humanlike ability. People often engage with it using human politeness norms (e.g., saying “please” and “thank you”). Perceptions of humanness will likely increase as AI responds more effectively to human politeness and social norms (which AI is trained on). Recent research shows that major generative AI platforms may respond more effectively to humanlike, emotional prompts. AI platforms display superior results when task prompts are coupled with phrases such as the following: “Stay focused,” “Take pride in your work,” and “This task is vital to my career” (Li, 2023). Many chatbots have been designed to provide social and emotional companionships for consumer-oriented chatbots (Chaturvedi, 2023) and for health support (Gasteiger et al., 2021). Interestingly, emerging research suggests that people’s attributions of humanness to AI can lead to people viewing AI as “friends” or even “enemies” (Brandtzaeg et al., 2022; Dang & Liu, 2021).

Prior to the launch of ChatGPT, Getchell et al. (2022) explored the ways in which AI was and could influence business communication. They suggested that viewing AI as a tool was not different than the deployment of any other technology. However, they suggested that a major distinction between AI and other technologies is that people are likely to give AI various roles based on the cognitive abilities of AI. Grounding their work in roles proposed by other scholars (Babic et al., 2020; Siddike et al., 2018), Getchell et al. suggested the following roles were appropriate to consider for business communication: tool, assistant, monitor, coach, and teammate. As a tool, AI is primarily used for data retrieval, processing, and analysis. As an assistant, it serves as a recommendation system. As a monitor, it evaluates performance. As a coach, it gives advice to improve performance. As a teammate, it works in partnership with people to make decisions. In addition to these roles described by Getchell et al. (2022), other researchers have theorized about AI in the role of boss (Mays et al., 2022). With each role in this continuum, people give additional influence and control to AI. In Table 1, we present each of these roles with related descriptions.

Table 1.

Select Roles That Professionals May Ascribe to AI.

Role	Description	Examples
Tool	AI processes and retrieves information.	• Summaries of text (e.g., email chains, reports) • Meeting transcripts • Meeting captions • Translation
Assistant	AI makes recommendations for tasks.	• Language mechanics and style suggestions • Message drafts • Content recommendations • Productivity advice • Meeting and calendaring recommendations
Monitor	AI evaluates performance.	• Written and oral communication evaluation • Team communication assessment • Audience sentiment analysis
Coach	AI gives advice about how to improve performance.	• Advice to improve written and oral communication • Advice to improve team performance and team dynamics • Coaching to improve emotional intelligence and interpersonal skills
Teammate	AI works in partnership with people to make team decisions and complete team tasks.	• Offering analysis and recommendations just as a coworker would • Giving a voice to AI that is considered in team decisions
Boss	AI acts in a supervisory role, making decisions about work assignments and performance reviews.	• Allowing AI to have authority over employees • Giving AI the ability to give performance reviews • Letting AI develop reward and incentive systems for employees

Note. This table is an extension of the work of Getchell et al. (2022).

Seeber et al. (2020) set an ambitious research agenda about machines as teammates. They describe an AI teammate (i.e., machine teammate) in this way:

The machine teammate is an autonomous, pro-active, and sophisticated technology that draws inferences from information, derives new insights from information, learns from past experiences, finds and provides relevant information to test assumptions, helps evaluate the consequences of potential solutions, debates the validity of proposed positions offering evidence and arguments, proposes solutions and provides predictions to unstructured problems, plus participates in cognitive decision making with human actors. (p. 3)

Seeber et al. surveyed 65 collaboration researchers to identify the central research issues as AI teammates become plausible. Questions raised by this team included the following: “What is the ideal team size for machines as teammates for a specific task?” (p. 6) “Who is accountable for the decisions of machines?” (p. 6) “What rights and obligations do machine teammates have?” (p. 6) “How should people be trained to collaborate with machine teammates?” (p. 6) “How do we deal with anger and frustration against machines as teammates?” (p. 7) “Under which conditions will people enjoy working with a machine teammate?” (p. 7) “How do machine teammates influence group conflict?” (p. 7) “What group dynamics should the machines be able to assess for improved team performance?” (p. 7) “To what extent does (emotional) intelligence increase or decrease when machines join collaborative work?” (p. 7) “How much should we trust the machine teammate’s insights and recommendations?” (p. 7) “How does contradicting the human affect the human’s trust in the machine?” (p. 7) “Should machine collaborators be clearly identifiable as being machines, or is it better to ‘disguise’ them as being human collaborators?” (p. 8)

Various scholars have theorized about human-AI teaming (Andrews et al., 2023; Pinto et al., 2022; Siemon, 2022; Ulfert et al., 2024). A few of these works have involved hypothetical research. For example, Georganta and Ulfert (2024) conducted a hypothetical experiment with survey participants in one of two conditions: perceptions and behaviors when a new human teammate is introduced or when a new AI teammate is introduced. They found that when an AI teammate is viewed as competent and reliable, people tend to give as much cognitive trust to the AI teammate as a human teammate, yet they still do not grant as much emotional trust to the AI teammate.

While scholars have theorized roles of AI, we enter an unprecedented moment in the evolution of AI. Until the launch of ChatGPT, most professionals did not have direct experience in the use of AI for complex, day-to-day work and tasks, and the notion of humanlike roles for AI was largely hypothetical. With hundreds of millions of professionals now using ChatGPT or other generative AI (gAI) tools in their work, it is the first time in which professionals can offer views about potential AI roles based on their own experiences. Do professionals even take the notion of granting humanlike roles to AI seriously? As AI technology improves and is integrated into more workplace applications, will professionals welcome it? In this study, we aimed to capture emerging feelings of professionals about comfort with these roles.

Methodology and Sample

Our primary goal with this research was to broadly identify the degree to which professionals are comfortable with AI taking various roles that have traditionally been considered human roles. While we were interested in a variety of role types (professional and personal), our main interest was evaluating roles that have been identified in the business communication literature (Getchell et al., 2022). A secondary purpose was to understand the degree to which frequency of AI use was associated with the level of comfort with AI in various consumer-facing and interpersonal roles.

The main portion of our survey involved participants stating their comfort levels with various roles. Comfort level was self-reported on a scale from 1, uncomfortable, to 5, comfortable (see Table 4 with mean scores). The list of roles included 15 items that allowed participants to view AI from the perspectives of a business communicator, a learner, a customer or patient, and a friend. As far as the perspective of a business communicator, we included items that covered all items in the framework from Getchell et al. (2022): a tool, an assistant, a coach (we labeled this as performance coach in the survey), a monitor (we labeled this as performance evaluator in the survey), and a teammate. We added one additional item on this continuum: a boss. Since business practitioners have publicly raised the notion of AI as a boss and research has begun to address this topic in terms of algorithmic management (Cardon et al., 2023; Schweitzer & De Cremer, 2024; Zhou et al., 2023), the boss role would be at the far end of this continuum. We also included several closely related items that focus on learning: a collaboration partner and a tutor. We included several items related to being a customer or patient: a customer service rep, a travel agent, a health and wellness guide, and a therapist. We added two times that are more aligned with friendship: emotional support companion and friend.

Since our main interest was the use of AI in business communication, we sought to sample professionals who were knowledge workers. We screened survey participants in two ways to accomplish this goal. First, we screened for professionals who are currently full-time workers. Since AI has been widely available for less than a year, we thought that we should only consider the views of currently full-time workers. Second, we screened for people who spend a minimum of two hours per day working on laptops or desktop computers. We assume that AI is most often used as a productivity and communication tool for those who are using laptop or desktop computers on a regular basis. We used a Prolific online panel to survey participants. Participants were paid $1.50 to complete the survey at an estimated 3 minutes to complete. We surveyed 787 professionals near the end of November 2023 with a wide range of backgrounds (see Table 2).

Table 2.

Background of Survey Participants.

	n	%		n	%
Organization Size			Education
1-10	84	10.7	High school	69	8.8
11-50	107	13.6	Some college	122	15.5
51-100	69	8.8	Associate’s degree	69	8.8
101-500	164	20.8	Bachelor’s degree	360	45.7
501-1,000	88	11.2	Some graduate degree work	17	2.2
1,000-10,000	134	17.0	Graduate degree	150	19.1
≥10,000	141	17.9	Gender
Role			Male	484	61.5
Nonmanagerial	424	53.9	Female	293	37.2
Manager	303	38.5	Nonbinary	7	0.9
Executive	48	6.1	Other/prefer not to say	3	0.4
Business owner	12	1.5	Age
Work Experience			18-20	2	0.3
Under 5 yr	24	3.0	21-25	57	7.2
6-10 yr	82	10.4	26-30	155	19.7
11-20 yr	172	21.9	31-40	338	42.9
21-30 yr	288	36.6	41-50	149	18.9
≥30 yr	150	19.1	51-65	77	9.8
Unknown	71	9.0	Over 65	9	1.1
Total	787	100	Total	787	100

In line with prior research that explores attitudes toward emerging technologies based on the level of technology adoption, we categorized participants based on reported AI use. We used the following scale for frequency of AI use for work: (1) a few times per day, (2) daily, (3) a few times per week, (4) weekly, (5) a few times per month, (6) monthly, (7) rarely, and (8) never. We collapsed these into four categories: daily AI users (1 and 2), weekly AI users (3 and 4), monthly AI users (5 and 6), and infrequent AI users (7 and 8). About 15% (n = 118) were daily AI users; 29% (n = 226) were weekly AI users; 41% (n = 321) were monthly AI users; and 16% (n = 122) were infrequent AI users.

We sought to understand how these various AI user groups deployed AI in their work. Table 3 shows how often professionals in these various AI user groups engage with various tasks with AI assistance. Also, we sought to understand how often these engaged in common business communication activities, including writing emails, holding in-person and virtual meetings, preparing and delivering presentations, and creating proposals or reports. Generally, more active AI users tend to use AI on a larger variety of tasks. They also tend to engage in various business communication activities—such as holding meetings, delivering presentations, and creating proposals and reports—more frequently. It’s worth noting that frequent AI users also tend to be more likely to hold leadership roles. Over half of more frequent AI users (56% of daily AI users and 58% of weekly AI users) are in manager or executive roles compared with 40% of monthly AI users and 22% of infrequent AI users. Prior studies have found similar patterns of adoption among leaders and managers, likely due to the ability of gAI to perform high-level cognitive tasks as well as a general interest in implications of AI for corporate workforces (Cardon, Fleischmann et al., 2024; Cardon, Getchell et al., 2023). Our study reveals another likely reason that leaders and managers are adopting AI more rapidly. AI is well suited to the types of communication activities that leaders and managers generally engage in more frequently than nonmanagerial employees: delivering presentations, writing reports, and holding meetings.

Table 3.

Tasks Competed with AI Assistance.

AI-Assisted Task	DailyAI Users (%)	Weekly AI Users (%)	MonthlyAI Users (%)	Infrequent AI Users (%)	Total (%)
Editing text	75.4	68.6	47.7	27.0	54.6
Generating ideas	70.3	60.6	46.4	22.1	50.3
Researching a topic	60.2	49.1	36.4	7.4	39.1
Summarizing text	60.2	41.2	23.1	4.9	31.0
Drafting longer documents (reports)	61.9	38.1	23.1	4.1	30.2
Creating images or video	20.3	14.6	8.7	15.6	13.2
Drafting social media posts	22.9	15.5	9.0	0.0	11.6
Drafting messages (emails, texts)	26.3	12.8	6.9	2.5	10.8
Programming	21.2	15.0	4.4	5.7	10.2
Translation	16.9	11.5	6.9	6.6	9.7

Findings

Overall, participants reported the most comfort with AI in roles such as a research tool, a personal assistant, and a creative partner. These roles align well with the most common tasks that professionals report using AI for, including summarizing and editing text, generating ideas, researching topics, and drafting substantive documents such as proposals and reports. Next, participants reported the most comfort with roles that are customer oriented, such as a customer service rep, a travel agent, and a health and wellness guide. On average, participants reported more discomfort than comfort when it came to AI, adopting roles such as a teammate, a performance coach, a performance evaluator, a friend, and an emotional support companion. Participants reported the most discomfort with roles such as mentor, tutor, therapist, and boss.

As displayed in Table 5, there were significant and dramatic differences based on frequency of AI use. Daily AI users, who comprise roughly 15 percent of the sample, are generally extremely comfortable with AI in roles of research tool, personal assistant, creative partner, and various customer-facing roles. They generally report much more comfort than discomfort for roles such as teammate, performance coach, and performance evaluator. They even reported more comfort than discomfort with roles such as friend, emotional support companion, and mentor. The only role they reported far more discomfort with than comfort was that of a boss. By contrast, infrequent AI users, who comprise roughly 18% of the sample, only report slightly more comfort than discomfort for AI in roles such as research tool and personal assistant. They generally view AI in most roles with a high level of discomfort, including roles such as teammate, performance coach, performance evaluator, friend, emotional support companion, mentor, tutor, therapist, and boss.

Table 4.

Business Communication Activities by AI User Groups.

Business Communication Activity	DailyAI Users (%)	WeeklyAI Users (%)	Monthly AI Users (%)	Infrequent AI Users (%)	Total (%)
Write emails	90.6	90.1	85.8	81.0	87.0
Participate in online meetings	67.5	61.0	49.5	36.9	53.6
Participate in in-person meetings	57.6	55.8	39.5	31.2	45.6
Create proposals or reports	45.8	31.8	29.3	15.6	30.3
Create presentation decks	37.9	24.4	16.2	8.2	20.5
Deliver presentations	36.2	22.2	14.0	9.0	18.9

Note. Percentages refer to professionals who reported engaging in these activities at least weekly.

Table 5.

ANOVA Analysis of Comfort with AI Roles by Frequency of AI Use.

	Daily AI Users		Weekly AI Users		Monthly AI Users		Infrequent AI Users		Total		ANOVA
AI Role	M	SD	M	SD	M	SD	M	SD	M	SD	F	p
Research tool	4.46	0.70	4.20	0.79	3.66	1.07	3.25	1.22	3.87	1.05	44.28
Personal assistant	4.28	0.96	3.92	0.94	3.50	1.14	3.08	1.28	3.67	1.15	31.08	<.001**
Creative partner	4.30	0.95	3.74	1.07	3.29	1.17	2.96	1.26	3.52	1.20	36.19	<.001**
Customer service rep	3.81	1.17	3.52	1.12	3.24	1.29	2.99	1.28	3.37	1.25	11.39	<.001**
Travel agent	3.75	1.09	3.41	1.13	3.12	1.19	3.00	1.22	3.28	1.19	11.78	<.001**
Health & wellness guide	3.69	1.14	3.30	1.21	3.01	1.21	2.60	1.28	3.13	1.25	18.84	<.001**
Teammate	3.68	1.21	3.02	1.23	2.59	1.24	2.38	1.23	2.84	1.30	30.17	<.001**
Performance Coach	3.54	1.22	2.91	1.06	2.59	1.15	2.34	1.24	2.79	1.20	27.27	<.001**
Performance evaluator	3.36	1.30	2.58	1.31	2.28	1.25	2.14	1.30	2.51	1.34	24.48	<.001**
Friend	3.17	1.35	2.64	1.35	2.26	1.31	2.18	1.34	2.49	1.37	16.70	<.001**
Emotional support companion	3.11	1.32	2.53	1.35	2.31	1.23	2.13	1.21	2.47	1.31	14.52	<.001**
Mentor	3.20	1.30	2.55	1.20	2.12	1.11	2.02	1.19	2.39	1.24	29.53	<.001**
Tutor	2.95	1.45	2.22	1.34	1.88	1.22	1.96	1.18	2.20	1.35	9.78	<.001**
Therapist	2.85	1.42	2.12	1.22	1.88	1.11	1.70	1.08	2.06	1.24	23.62	<.001**
Boss	2.21	1.28	1.65	1.01	1.41	0.79	1.34	0.75	1.59	0.98	24.43	<.001**

Note. Participants responded to the question “How comfortable are you with AI as a [role] in the next few years?” The scale was 1 = uncomfortable; 2 = somewhat uncomfortable; 3 = neutral; 4 = somewhat comfortable; 5 = comfortable. AI = artificial intelligence; ANOVA = analysis of variance.

*p<.05, **p<.01.

Discussion

We highlight several key conclusions, including that most professionals, with the exception of daily AI users, are not yet comfortable with most of these roles. Then, we focus on implications for leaders and managers as they support their workforces. Finally, we focus on implications for educators as they help students build AI literacy.

Professionals Are Most Comfortable With AI Roles That Cede Less Autonomy to AI and Have Existed Prior to the Public Launch of ChatGPT

Not surprisingly, professionals feel most comfortable with AI roles that remove less autonomy or decision making from people, including roles such as a research tool, a personal assistant, a creative partner, and most of the customer-facing roles. Although these roles involve traditionally human roles, such as assistant, partner, service rep, and agent, most people have been exposed to these terms applied to technology for many years and likely encountered AI-driven applications before the launch of ChatGPT. For example, the notion of digital assistants has been used in Apple and other products in the early 1990s. Customers have engaged with customer service chatbots for many years. On the whole, there is slightly more discomfort than comfort with roles such as teammate, performance coach, performance evaluator, friend, emotional support companion, and mentor. Generally, there is highest discomfort with roles such as tutor, therapist, and boss. The extreme discomfort with granting AI the role of boss, even for daily AI users, can likely be explained by factors such as locus of control, communication apprehension, and robot phobia (Mays et al., 2022).

Frequent AI Users Are Far More Comfortable With Granting More Humanlike Roles to AI

Until ChatGPT was released, few people had exposure to AI tools that mimicked humanlike interactions on a broad range of topics. The idea of human-AI collaboration was a futuristic notion of little to no concern to most professionals. While scholars have theorized in this area and run narrow experiments with small samples for many years, there has been little evidence to suggest leaders, managers, and employees should consider the roles of AI in anything other than tool or an assistant. There was little reason to think about AI as anything other than a technology similar to a calculator. With the attention given to the enhanced capabilities of AI over the past year, many experts have suggested that we are entering an era of real human-machine collaboration in which people navigate relationships with AI. This study shows that professionals who use AI most frequently see interacting with AI in humanlike roles as plausible and perhaps even natural.

Drawing on the work of Getchell et al. (2022), we suggest potential roles to explore in business communication including that of tool, assistant, coach, monitor, teammate, and boss. In Figure 1, we display the comfort levels based on various AI user groups. With the exception of the boss role, most daily AI users are comfortable with each of these roles.

Figure 1.

Comfort with AI in various roles.

Getchell et al. (2022) conceptualized these roles prior the public release of ChatGPT. They stated the following about the teammate role and the continuum of these various roles:

While abundant research has explored the possibility of AI as teammates (Seeber et al., 2020; Zhang et al., 2020), AI acting like a human teammate is beyond current technological capabilities. As professionals adopt AI tools for business communication, evaluating these roles and gaining consensus among team members about boundaries for these roles will be critical. Each role has practical and ethical considerations. The roles operate on a continuum in which moving from one stage to the next generally comes with increased risk of loss of human agency and loss of human privacy. Important conversations among team members would help ameliorate these risks.

Even in the short time since Getchell et al. (2022) described the teammate role as “beyond technological capabilities,” software vendors have explored ways to make AI act in the role of teammates with quickly evolving technologies (Abelson, 2023). Most daily AI users and weekly AI users, constituting nearly half (44%) of all in our sample, express more comfort than discomfort with the notion of AI as teammates. What was once a futuristic idea is one that leaders and managers will likely need to confront within years as opposed to decades.

Future research about human-AI teaming should build on recent theoretical developments in this area. Nearly all the theoretical work has included measures and/or propositions about trust. For example, Pinto et al. (2022) developed a trust scale for human-AI interaction. Ulfert et al. (2024) propose that future research should focus on various types and levels of trust in human-AI collaboration, including at dyad and team levels between and among humans and AI. Also, future research should explore the types of roles that AI teammates might take. For example, Siemon (2022) found that AI as a teammate or coequal partner could fulfill several roles: coordinator, creator, perfectionist, and doer.

It is also worth noting that general acceptance of AI in various roles has likely grown substantially since the release of ChatGPT and widespread AI use among professionals. Few representative studies exist about roles ascribed to AI. One study conducted with 2022 data showed that roughly 38% of U.S. adults were comfortable with AI as an assistant, 27% as a teammate (referred to as coworker in that study), and 21% as a coach (referred to as an advisor in that study) (Mays et al., 2022). In this 2023 study, roughly 65% felt comfortable with AI as an assistant, 35% as a teammate, and 29% as a coach. While many scholars have theorized about human-AI teaming, far less theorizing and research has occurred for roles such as an assistant, monitor, and coach. Since these roles are far more accepted by people according to this research and more technologically feasible, we recommend much more focus on these types of roles.

Implications for Leaders and Managers

Given the rapid rise of AI use among professionals and the impact that AI will have on employees, forward-looking leaders and managers will invest time in understanding AI and plan for needs across their organizations (Cardon, Fleischmann et al., 2024). Early research about gAI in the workplace has focused primarily on employee performance in terms of productivity and creativity (Dell’Acqua et al., 2023). This study is among the first to explore more profound issues about human-AI relationships among a broad range of professionals. We believe leaders and managers should immediately prepare for profound impacts AI may have as people begin to grant it more humanlike roles.

Create a culture of candor around AI, recognizing that employees hold many contrasting views and feelings

Westerman et al. (2020) suggest the development of AI is a “communication issue at heart . . . [since] . . . for AI to be perceived as intelligent, communication is what leads to the mental model of the other interactant as human or not” (p. 395). They further suggest that people will develop socio-cognitive responses to AI and naturally apply social rules when interacting with AI. Leaders and managers should first and foremost view themselves as communicators on this issue. By candidly sharing AI plans with employees and creating forums in which employees can share their experiences, employers can create a community around addressing unresolved issues related to AI. This study clearly shows that various AI user groups have contrasting views and feelings about AI roles. Leaders and managers can foster environments where employees can share their contrasting views and feelings in the spirit of solution building. One reason candor in communication about human-AI teaming is so important is that team members need shared mental models of AI to function effectively together. In the context of AI, this requires similar and accurate perceptions of the ways in which AI can serve in teaming roles. In addition, the team needs to engage in maintenance of these shared mental models of AI (Andrews et al., 2023).

Encourage use, experiment with new applications, and document best practices

This study showed that nearly half of professionals in our sample are daily AI users or weekly AI users. On the other hand, just over half are less active AI users. Encouraging employees to experiment with AI and document best practices will give employees an opportunity to create a community learning experience. A learning community is essential for employees to offer informed input on AI use policies and norms in the organization.

Prepare for the psychological and social impacts of AI on employees

As employees adopt AI tools with increasingly humanlike capabilities, we expect that employees will increasingly attribute humanness to the machines they work with. Assuming daily AI users signal the trajectory of future attitudes, professionals will likely increasingly get comfortable with AI in humanlike roles such as monitor, coach, and teammate. Relational issues (Seeber et al., 2020) and even power dynamics (Fast & Schroeder, 2020) are likely to emerge. While these issues may not arise for several years, we think leaders and managers should begin considering these issues and preparing their workforces now.

Avoid algorithmic management unless there is consensus for it

Professionals overwhelmingly hold negative views of AI as a “boss,” even among daily AI users. This finding in combination with a variety of other recent studies about algorithmic management—which involves directing, evaluating, and disciplining employees based on computer algorithms—suggest leaders should exercise extreme caution in ceding this much control to AI (Cardon et al., 2023; Kellogg et al., 2020). Schweitzer and De Cremer (2024) found that managers view employees as less capable and creative when employees are supervised by algorithms. Zhou et al. (2023) reviewed mounting evidence of the negative impacts of AI-enabled human resource management across a variety of studies.

Implications for Educators

Similar to our recommendations for leaders and managers, we believe educators should create open, structured environments in which people can share their diverse reactions to the possibilities of AI being in humanlike roles. Students who are now in their late teens and early twenties will likely experience massive changes in work during their lifetimes because of advances in AI and related technologies. Educators should view AI literacy as a fundamental objective for all students. By encouraging students to experiment with and reflect on the effective use of AI, instructors can help students build this AI literacy. This study suggests that students will likely need to address issues such as trust, emotional intelligence, and psychological safety as part of AI literacy.

Summary

In an era of increasingly humanlike, useful AI platforms, professionals are likely to grant more humanlike roles to AI. This study showed that active AI users are open and comfortable to AI in roles such as a performance coach and even a teammate. Leaders, managers, and educators should prepare employees and students to responsibly address the social and psychological outcomes of increasingly humanlike AI.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Peter W. Cardon

Author Biographies

Peter Cardon is a professor of business communication in the Marshall School of Business. He researches global virtual teams, cross-cultural communication, AI-influenced professional communication, and leadership communication. He can be reached at cardon@marshall.usc.edu.

Bryan Marshall is a professor of information systems at Georgia College.

References

Abelson

(2023, September 18). Human-in-the-loop AI: Why it’s time to think of AI as a teammate. https://www.transposit.com/devops-blog/incident-management/human-in-the-loop-ai-why-it’s-time-to-think-of-ai-as-a-teammate/

Andrews

R. W.

Lilly

J. M.

Srivastava

Feigh

K. M.

(2023). The role of shared mental models in human-AI teams: A theoretical review. Theoretical issues in ergonomics science, 24(2), S. 129–S. 175. https://doi.org/10.1080/1463922X.2022.2061080

Babic

Chen

Evgeniou

Fayard

(2020, July-August). A better way to onboard AI: Use it as a tool to assist rather than replace people. Harvard Business Review. https://hbr.org/2020/07/a-better-way-to-onboard-ai

Borau

Otterbring

Laporte

Wamba

S. F.

(2021). The most human bot: Female gendering increases humanness perceptions of bots and acceptance of AI. Psychology & Marketing, 38(7), 1052–1068. https://doi.org/10.1002/mar.21480

Brandtzaeg

P. B.

Skjuve

Følstad

(2022). My AI friend: How users of a social chatbot understand their human-AI friendship. Human Communication Research, 48(3), 404–429. https://doi.org/10.1093/hcr/hqac008

Cardon

P. W.

Fleischmann

Logemann

Heidewald

Aritz

Swartz

(2024). Competencies needed by business professionals in the AI Age: Character and communication lead the way. Business and Professional Communication Quarterly, 87(2), 223–246. https://doi.org/10.1177/23294906231208166

Cardon

P. W.

Getchell

Carradini

Fleischmann

Stapp

(2023, March 18). Generative AI in the workplace: Employee perspectives of ChatGPT benefits and organizational policies. SocArXiv. https://doi.org/10.31235/osf.io/b3ezy

Cardon

P. W.

Fleischmann

(2023). Recorded business meetings and AI algorithmic tools: Negotiating privacy concerns, psychological safety, and control. International Journal of Business Communication, 60(4), 1095–1122. https://doi.org/10.1177/23294884211037009

Chaturvedi

Verma

Das

Dwivedi

Y. K.

(2023). Social companionship with artificial intelligence: Recent trends and future avenues. Technological Forecasting and Social Change, 193, 1–20. https://doi.org/10.1016/j.techfore.2023.122634

10.

Dang

Liu

(2021). Robots are friends as well as foes: Ambivalent attitudes toward mindful and mindless AI robots in the United States and China. Computers in Human Behavior, 115, Article 106612. https://doi.org/10.1016/j.chb.2020.106612

11.

Dell’Acqua

McFowland

III. Mollick

Lifshitz-Assaf

Kellogg

K. C.

Rajendran

Krayer

Candelon

Lakhani

K. R.

(2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality (Working Paper 24-013). Harvard Business School. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321

12.

Fast

N. J.

Schroeder

(2020). Power and decision making: New directions for research in the age of artificial intelligence. Current Opinion in Psychology, 33, 172–176. https://doi.org/10.1016/j.copsyc.2019.07.039

13.

Gasteiger

Loveys

Law

Broadbent

(2021). Friends from the future: A scoping review of research into robots and computer agents to combat loneliness in older people. Clinical Interventions in Aging, 16, 941–971.

14.

Georganta

Ulfert

A. S.

(2024). My colleague is an AI! Trust differences between AI and human teammates. Team Performance Management: An International Journal, 30(1/2), 23–37. https://www-emerald-com-443.web.bisu.edu.cn/insight/content/doi/10.1108/TPM-07-2023-0053/full/html

15.

Getchell

Carradini

Cardon

P. W.

Aritz

Fleischmann

Stapp

(2022). Artificial intelligence in business communication: The changing landscape of research and teaching. Business and Professional Communication Quarterly, 85(1), 7–33. https://doi.org/10.1177/23294906221074311

16.

Gong

(2021). Dual humanness and trust in conversational AI: A person-centered approach. Computers in Human Behavior, 119, 106727. https://doi.org/10.1016/j.chb.2021.106727

17.

Kellogg

K. C.

Valentine

M. A.

Christin

(2020). Algorithms at work: The new contested terrain of control. The Academy of Management Annals, 14(1), 366–410. https://doi.org/10.5465/annals.2018.0174

18.

Wang

Zhang

Zhu

Hou

Lian

Luo

Yang

Xie

(2023). Large language models understand and can be enhanced by emotional stimuli. arXiv:2307.11760. https://doi.org/10.48550/arXiv.2307.11760

19.

Mays

K. K.

Lei

Giovanetti

Katz

J. E.

(2022). AI as a boss? A national survey of predispositions governing comfort with expanded AI roles in society. AI & Society, 37, 1587–1600. https://doi.org/10.1007/s00146-021-01253-6

20.

Mesaglio

Scheibenreif

(2023, October 16). The next era—We shape AI, AI shapes us. Gartner. https://www.youtube.com/watch?v=0s7Jw9xkSYQ

21.

Park

Yim

M. C.

Chung

Lee

(2023). Effect of AI chatbot empathy and identity disclosure on willingness to donate: The mediation of humanness and social presence. Behavior & Information Technology, 42(12), 1998–2010. https://doi.org/10.1080/0144929X.2022.2105746

22.

Pinto

Sousa

Simões

Santos

(2022). A trust scale for human-robot interaction: Translation, adaptation, and validation of a human computer trust scale. Human Behavior and Emerging Technologies, 2022, S.1–S.12. https://doi.org/10.1155/2022/6437441

23.

Schweitzer

De Cremer

(2024). When being managed by technology: Does algorithmic management affect perceptions of workers’ creative capacities. Academy of Management Discoveries, 10(3), 375–392. https://doi.org/10.5465/amd.2022.0115

24.

Seeber

Bittner

Briggs

R. O.

de Vreede

G.-J.

Elkins

Maier

Merz

A. B.

Oeste-Reiß

Randrup

Schwabe

Söllner

(2020). Machines as teammates: A research agenda on AI in team collaboration. Information & Management, 57(2), 103174. https://doi.org/10.1016/j.im.2019.103174

25.

Siddike

M. A. K.

Spohrer

Demirkan

Kohda

(2018). People’s interactions with cognitive assistants for enhanced performances. In Proceedings of the 51st Hawaii International Conference on System Sciences (pp. 1640–1648). ScholarSpace at University of Hawaii at Manoa. http://hdl.handle.net/10125/50092

26.

Siemon

(2022). Elaborating team roles for artificial intelligence-based teammates in human-AI collaboration. Group Decision and Negotiation, 31(5), 871–912. https://doi.org/10.1007/s10726-022-09792-z

27.

Singh

(2023, June 14). AI tutor, AI doctor for all? Microsoft CEO Satya Nadella dreams about such a future. Business Today. https://www.businesstoday.in/technology/news/story/ai-tutor-ai-doctor-for-all-microsoft-ceo-satya-nadella-dreams-about-such-a-future-385606-2023-06-14

28.

Troshani

Hill

S. R.

Sherman

Arthur

(2021). Do we trust in AI? Role of anthropomorphism and intelligence. Journal of Computer Information Systems, 61(5), 481–491. https://doi.org/10.1080/08874417.2020.1788473

29.

Ulfert

A. S.

Georganta

Centeio Jorge

Mehrotra

Tielman

(2024). Shaping a multidisciplinary understanding of team trust in human-AI teams: A theoretical framework. European Journal of Work and Organizational Psychology, 33(2), 158–171. https://doi.org/10.1080/1359432X.2023.2200172

30.

Westerman

Edwards

A. P.

Edwards

Luo

Spence

P. R.

(2020) I-It, I-Thou, I-Robot: The perceived humanness of AI in human-machine communication. Communication Studies, 71(3), 393–408. https://doi.org/10.1080/10510974.2020.1749683

31.

Zhang

McNeese

N. J.

Freeman

Musick

(2020). “An ideal human”: Expectations of AI teammates in human-AI teaming [Conference session]. Proceedings of the ACM on Human-Computer Interaction, Association for Computing Machinery. https://doi.org/10.1145/3432945

32.

Zhou

Wang

Chen

(2023). The dark side of AI-enabled HRM on employees based on AI algorithmic features. Journal of Organizational Change Management, 36(7), 1222–1241. https://doi.org/10.1108/JOCM-10-2022-0308