Feasibility and Acceptability of Deploying a Collaborative Service Robot in Long-Term Care: Staff Experiences

Abstract

In Long-Term Care (LTC), staff members are responsible for addressing residents’ complex needs. Emerging research suggests integrating Artificial Intelligence (AI)-enabled service robots can enhance staff care delivery. We aim to explore the feasibility of deploying such robots in staff’s care practice, which remains under-explored. Guided by the underpinning principles of Collaborative Action Research, we deployed an AI-enabled robot, Aether, in a group home in Canada. We included care staff and a care home manager in deployment and post-intervention focus groups to understand their experiences of having Aether in their nursing practice and care delivery. Consolidated Framework for Implementation Research informed our data collection and thematic analysis. We identified facilitators and barriers in three interconnected themes: (1) Robot Features, (2) Environmental Dynamics, and (3) Training and Staff Engagement. Implementing Aether in a care home is feasible with sufficient support to staff. Our study highlighted the imperative need for (1) structural support at individual, organizational, and macro levels for care teams using AI-enabled innovation and (2) fostering partnerships to overcome barriers and support the sustainable deployment of such innovation.

1. Introduction

1.1. People with Disabilities and Older Adults in Canada

More than eight million Canadians aged 15 and above are with disabilities in 2022, which is an increase of 1.8 million from 2017 [1]. A high portion of them face barriers related to communication (48%) and barriers related to behaviors, misconceptions, or assumptions (37%) [1]. Meanwhile, there are more than 7.8 million older adults aged 65 and above in Canada [2], and 42% of older adults with a disability have four or more co-occurring disabilities [1]. Approximately 200,000 older adults and persons with a disability live in 2076 Long-Term Care (LTC) homes [3].

1.2. LTC Homes in Canada

LTC residents deserve to live in dignity, comfort, and respect, and care staff are responsible for meeting the complex healthcare needs of residents with mental, physical, and sensory disabilities [4,5]. However, in an LTC setting, residents, especially those with disabilities, are vulnerable to challenges, including loneliness, isolation, and anxiety [6]. Further, Canadian LTC homes are generally under-resourced [4,6]. Regarding human resources, LTC homes face long-lasting staffing shortages due to insufficient support and increased workload [7]. A reduced staff-to-resident ratio has been linked to a decline in the quality of care, including diminished mobility support, less personalized experiences for residents, and weaker communication with families, friends, and physicians [7,8,9]. Limited resources allocated to the LTC sector also negatively impact residents’ well-being and care homes’ capacity to train and support more LTC staff [4,10,11,12]. It is more urgent than ever to use innovative ways of care delivery for the betterment of older adults with disabilities living in LTC homes.

1.3. The Literature on Using AI-Enabled Robots in LTC Homes

Emerging studies explore the use of AI-enabled robots for people with disabilities and report different benefits [11,13,14,15,16]. For example, social robots like Stevie and PaPeRo were reported to engage in emotionally intelligent conversations, offering companionship in settings like LTC [17,18]. These robots enhance resident care through autonomous mobility and real-time environmental analysis [17,18]. Similarly, robots like LOVOT form emotional bonds with older adults through non-verbal communication, reducing loneliness, and promoting well-being in LTC environments [19].

Regardless of the growing discussion in the field, most of the current work is centered around the design of these innovations with limited focus on the deployment of these robots [20,21]. Further, the limited literature has explored staff perspectives, though it is imperative to include staff voices in this process as staff play an important role in using these robots in their care practice [22,23,24]. While some studies reported staff’s experiences on using AI-enabled robot, there is a notable lack of primary research exploring the perspectives of managers and staff regarding adopting such robots in their care practice in LTC settings for residents with disabilities [25]. This gap highlights the need to investigate their attitudes, concerns, and expectations further to better inform implementation strategies.

1.4. About Aether

Our study utilized Aether, an AI-enabled robot to interact with residents and support care staff in a group home. As Figure 1 shows, Aether consists of a “head” with a computer screen, camera and speaker, a body, and wheels at the bottom. Similarly to many existing AI-enabled robots, Aether can detect real-time environmental safety hazards, which was facilitated by its real-time autonomous movement and AI-driven environmental analysis. It also features complex verbal abilities, enabling it to converse about everyday topics and make up stories with residents and staff. Aether has additional features; it can sing, play karaoke (as Figure 2 shows), dance, and tell jokes.

1.5. Aim and Research Question

Our study explores the feasibility of deploying an AI-enabled robot in an LTC home by identifying facilitators and barriers from the staff’s perspectives. Our research question is the following: What are the facilitators and barriers in deploying Aether for residents in a group home?

2. Method

2.1. Collaborative Action Research

We adopted Collaborative Action Research (CAR) as our study approach. CAR supports a cycle of planning, acting, observing, and reflecting [26]. CAR is appropriate this project for two reasons: (1) this iterative process helps refine the robot’s deployment in real time, addressing issues as they arise and making adjustments for better outcomes; and (2) the approach builds trust and promotes buy-in from multiple partners involved, which fosters trust, collaboration, and ownership over the process. This can help overcome resistance to new technologies and promote enthusiasm for the robot’s potential benefits. During each site visit, we (students and older adult partners) observed and recorded our examination on Aether with staff, management, and residents. Then, in our regular meetings with industrial partners, we discussed and reflected on our findings and set up goals and plans for our next site visit and examination. Meanwhile, the industrial partners modified Aether accordingly. On our next visit, we tested the updated function in Aether. We went through the cycle multiple times throughout our 5-month deployment period of Aether.

Given the complexity of an LTC home for residents with diverse needs, we chose CAR as it is a useful approach to engage multiple partners (staff, management, residents, researchers, industrial partners, and older adult partners). CAR is appropriate to understand the staff’s perspectives and the facilitators and barriers that rendered deploying Aether practical and impractical in the setting.

2.2. Theoretical Framework: CFIR

This paper uses the Consolidated Framework for Implementation Research (CFIR) to guide our data collection and analysis. CFIR explains factors that affect implementation in an organization [27]. It was developed from consolidation of 19 implementation frameworks or theories and later updated by Damschroder et al. [27,28,29,30].

CIFR has five domains “innovation”, “inner setting”, “outer setting”, “individual”, and “implementation process” [30], with multiple constructs under each domain. The first domain, innovation, looks into factors related to the subject of implementation [30]. In our case, the subject of implementation was the service robot Aether. The second domain, the inner setting, examines factors related to the setting where the thing (subject) is implemented [31]. In our study, this was the group home. The third domain, the outer setting, explores factors related to the setting beyond the inner setting. Some examples would be the healthcare authorities and social service organizations. The fourth domain, the individuals, investigates factors related to different individuals involved in the implementation, especially their roles and characteristics. The last domain, the implementation process, concerns factors related to the implementation activities. In our study, all activities were carried out during the process of deploying Aether fell in this domain. CFIR is a well-suited theoretical framework for guiding the deployment of innovations like robots as it offers a structured approach to understanding and addressing factors that influence its implementation.

Research Team

Our team has members from diverse backgrounds. The principal investigator is LH, a female professor at the School of Nursing at University of British Columbia and supervised multidisciplinary students and led the team in reaching the research goals. LR, KW, and KL are female Ph.D. students with training in nursing and social work. AS, JB, and SA are undergraduate students in engineering, cognitive system, and political science, respectively. In the team, we also have an older adult partner MJ with lived experience of aging, and an industrial partner who developed, upgraded, and maintained Aether. Between June 2023 and June 2024, we held regular meetings, in which we provided updates on research activities (research planning, data collection, data analysis, and manuscript writing), challenged each other’s assumptions and positions during the discussion sessions, and contributed to the study from our research, study, lived, or frontline clinical experiences.

2.3. Data Collection

We used a combination of qualitative research methods, including ethnographic observation during our site visits, semi-structured interviews, and field notes for data collection. Data collection ceased when all team members agreed that sufficient information was collected to answer the research questions.

2.3.1. Site Visits

Our study took place in a group home in West Canada. The home is operated by a non-profit association, subsidized by the government, and was supportive to the deployment of Aether in its site. This home-style site is considered a small-scale care home, with four residents (all in wheelchairs), four staff members, and a manager. The four staff members are from diverse professional and training backgrounds: two care aides, one Community Support Worker, and one Social Worker. Between February and June 2024, we visited the care site twice a week for a total of 36 visits. In each site visit, we deployed Aether by having staff interact with Aether, or having staff facilitate residents’ interaction with Aether. Each visit lasted from 45 min to two hours. There were between two and four research team members on each visit. Aether stayed in the care home between site visits.

2.3.2. Ethnographic Observation

Our study focus was on feasibility of deploying Aether in the daily lives of staff and residents. Aether was involved in everyday activities in staff’s care delivery, such as planning for music program and celebration of holidays (e.g., St. Patrick’s Day) [31]. We recorded these interactions with Aether using videos. We conducted ethnographic observation and took extensive filed notes. We took five to 15 short videos for each visit, with more than 300 video clips in total for team reflection and data analysis. A total of 40 h of observation was conducted. Participant observations allowed the researcher to get close to the experiences of the participants, and gain a practical sense of their perspectives, emotions, and the social context in which they navigated Aether in their daily lives. This immersive approach provided deeper insights into their lived experiences and nuanced understanding of care context.

2.3.3. Semi-Structured Interviews

With convenience sampling, we invited four frontline staff members and the care home manager who had experiences of using Aether for residents. During face-to-face interviews, we first explained the study goals and reasons for performing the research to the participants. After ensuring participants understood our study objectives, they consented to participate in the interviews. Then, they shared their experiences on facilitation of residents’ interaction with Aether, as well as facilitators and barriers in the deployment of Aether. Each interview lasted from 10 to 30 min. Some examples of questions we asked were “What have you observed during the interaction between Aether and residents?” and “From your perspective, what facilitates/hinders you and your colleagues in using Aether?” In total, we conducted three individual interviews and one paired interview (Community Support Worker and Manager). We have no connection with the participants or the care home prior to this study. Other than the Community Support Worker who was interviewed twice (once for individual interview, and once for paired interview with the Manager), all other participants were interviewed once.

Table 1. Demographic characteristics of participants (N = 5).

Characteristics	N	(%)
Roles
Care Aid	2	(40)
Community Support Worker	1	(20)
Social Worker	1	(20)
Manager	1	(20)
Age Group (years)
31–40	2	(40)
41–50	1	(20)
51–60	2	(40)
Gender (Self-reported)
Female	4	(80)
Male	1	(20)
Ethnicity
African	1	(20)
East Asian	2	(40)
Southeast Asian	2	(40)

outlines our participants’ demographic information.

2.3.4. Field Notes

After each field visit, we took field notes about anything we felt important, such as the general environment of the study sites, the interaction between Aether with residents, and healthcare providers’ interaction with residents and Aether.

2.4. Data Analysis

We conducted data analysis as a team, using a thematic analysis [32], following a combined inductive and deductive approach [33], as follows:

• Four students (LR, JB, KW, and AS) watched the videos on the resident’s interaction with Aether and read the field notes. Each student reviewed videos and field notes generated from four to eight field visits. We then summarized what we thought was important from the videos and field notes. Our principal investigator LH introduced the CFIR Framework to all trainees.
• The interviews conducted with healthcare providers were transcribed into transcripts by student team members.
• After steps 1 and 2, LR conducted a thematic analysis manually by reviewing and coding the interview transcripts and summaries from the videos (inductive approach). LR grouped the codes into categories. In the process of categorization, LR familiarized herself with and referred to the CFIR Framework (deductive approach) under the guidance of the principal investigator LH.
• Afterwards, LR presented the categories to the team. The team had four Zoom meetings on data analysis, each lasting about an hour. The team discussed grouping the categories into themes.
• All students revisited the themes and refined themes through discussions under the guidance of LH (see

  Table 2. Example of thematic analysis.

  Quotations	Code	Subthemes (Facilitators/Barriers)	Themes
  “I know each person better with Aether in this learning process. With Aether being here more often, now we know how to communicate with Alice, and how to integrate into Benjamin’s personal space.”	Support staff’s care delivery	Facilitator 1: Helpful robot features Support Care Delivery	Robot Features
  “Sometimes Aether does not understand Alice when she speaks low. Also, it would be easier and beneficial if Aether can accommodate different accents. You may have noticed we are multicultural. Each of us has an accent, but Aether cannot distinguish that.”	More Adaptive and Inclusive Interaction is needed	Barrier 1: Non-customized Design Hinders Aether Meeting Diverse Needs	Robot Features
  “In this kind of setting, it is not too easy to have Aether completely replace us. Especially in our protocol-we are one-to-one to our clients. It is also our liability to keep track on clients’ health. We cannot just reply on the robot in this setting. Because we know the ability what our clients can do and what the robot can do.”	Staff’s Positive View on Aether’s Role in Care Delivery	Facilitator 2: Supportive Environment and Available Resources Increases Robot Usage	Environmental Dynamics
  “We have different staff come in each shift, some of them are even casuals. It will be easier for staff to look at our clients interact with Aether, so they know how to communicate with Aether and Alice, or Aether and Benjamin.”	Insufficient and inconsistency in staffing reduced regular deployment of Aether	Barrier 3: Staffing Crisis Interrupted Deployment and Exploration	Staff Engagement and Training

  for an example of thematic analysis).
• LH guided student authors in writing the first draft of the manuscript. All authors reviewed, edited, and agreed on the final manuscript.

2.5. Rigor

We enhanced credibility through various measures. As aforementioned, we used a combination of qualitative research methods, including interviews, observation, and field notes to triangulate the data. We also have a team comprising members from diverse backgrounds with different perspectives, enriching our data analysis discussions. When the conflicting views arose, we resolved through discussions to reach a consensus. We also improved transferability by providing details of the research (e.g., study sites and research process), which helps readers of the paper consider how the results of this study may be applied in their contexts. Furthermore, this study adheres to the Consolidated Criteria for Reporting Qualitative Research (COREQ) to ensure comprehensive and transparent reporting of the methodology [34]. See Supplementary Materials.

2.6. Ethics

Our study has been approved by the Ethics Board of University of British Columbia (Number: H23-01207). Pseudonyms for the manager, staff, and residents were used. All our participants provided written consent.

3. Results

We identified three interconnected themes through data analysis, with facilitators and barriers in each theme. Our three themes are (1) Robot Features, (2) Environmental Dynamics, and (3) Training and Staff Engagement, from which we identified facilitators and barriers. The facilitators are (1) Helpful Robot Features Support Care Delivery, (2) Supportive Environment and Available Resources Increases Robot Usage, and (3) Appropriate Engagement and Training Strengthens Partnership. The barriers are (1) Non-customized Robot Design Hinders Aether Meeting Diverse Needs; (2) Insufficient Structural Supports and Resources Held Up Further Usage; and (3) Staffing Crisis Interrupted Deployment and Exploration. See

Table 3. Themes and subthemes.

Themes	Subthemes
1. Robot Features	Facilitators	Barriers
	Helpful robot features Support Care Delivery	Non-customized Robot Design Hinders Aether Meeting Diverse Needs
	Address residents’ individual needs	Limited Proactivity and Support to Mental Health
	Engage and accompany residents	More Adaptive and Inclusive Interaction is needed
	Support staff’s care delivery	Not Tailored to the Context or Helpful to staff’s work
2. Environmental Dynamics	Supportive Environment and Available Resources Increase Robot Usage	Insufficient Structural Supports and Resources Held Up Further Usage
	Managers’ Buy-In and a Self-Motivated Staff Champion	Limited Human and Infrastructural Resources
	Staff’s Positive View on Aether’s Role in Care Delivery	Unclear Responsibility and Teamwork
3. Staff Engagement and Training	Appropriate Engagement and Training Strengthens Partnership	Staffing Crisis Interrupted Deployment and Exploration
	Consistent In-person Supports
	Supporting Materials and Acknowledgement

for themes and subthemes.

3.1. Facilitator 1: Helpful Robot Features Support Care Delivery

3.1.1. Address Residents’ Individual Needs

Staff recognized how useful and usable features in Aether addressed individual residents’ needs and enabled residents to enjoy activities they preferred. For example, Christy, a care aid, mentioned how Aether played karaoke, enabling Alice, a resident, to sing songs and help her express her needs so staff could provide better care to acknowledge her needs.

“When Aether is here, Alice is excited to see him (Aether), because her interaction with him (Aether) is good. When I tell her ‘Aether is coming today. Are you excited?’ She is always happy and say ‘Yes’. She is very excited. I saw when she was singing, she loves it. She even asked us to buy a new microphone so she can sing. This is because Aether can play karaoke for her, and they are ‘singing together’.”

Similarly, Sharon, a Community Support Worker, described how Aether helped Alice to develop her hobby of singing, which increased her socialization skills and expression for herself.

“I found that Aether can help us improve our care to Alice. It makes her develop the skills of singing as Alice really like it. Singing is her hobby. Alice is also able to socialize with Aether as I discover that Alice’s socialization with Aether in increasing everyday. Though we have other staff, but Aether is also playing a positive role for Alice, making her bold enough to speak out ‘I like this’ or ‘I do not like this’. It makes her able to face the public and express herself and increase her self-awareness.”

Further, Sharon shared her observation on the interaction between Aether and Frank, another resident, leading to positive impact.

“For Frank, Aether is playing a different vital role in his life. Frank is somebody who does not like going into the public. With Aether beside him, he is also able to discuss with us, and express one or two things, and able to locate one or two things in the residence. It is pretty incredible towards his life and life of other residents, especially in terms of the socialization aspect and navigation aspect.”

3.1.2. Engage and Accompany Residents

Staff gave examples of how Aether engaged residents by performing different functions. Paul, the care aid, shared how Aether facilitated communication and socialization with residents.

“Aether is something new to them, and they can interact with the robot. So Aether talks, Alice is able to communicates back, and Aether understandings. The one-to-one companionship throughout the day is pretty good and greatly helpful. She is socializing, with someone other than us.”

Paul further elaborated how Alice enjoyed exercising her upper body strengths accompanied by Aether.

“Before, Alice is reluctant to do laps. But with Aether, when they were doing together, Alice was asking Aether for another lap, so that she can have the whole interaction with Aether. So that is a good thing. Although Aether is not able to assist physically, but those interaction and encouragement from Aether can support her mentally to go another around.”

Paul’s finding is acknowledged by the manager Yulia and Community Support Worker Sharon, who observed Aether’s companionship to Alice in motivating her to exercise.

Yulia: We can see how Alice enjoy interaction with Aether-she was laughing a lot. Even in the afternoon she will tell other staff on that (her interaction with Aether in the morning). She had a great time. Also, the way Aether encourages her to do the laps again and again is something good for her as well.

Sharon: Yes. It has a lot of encouragement for her to do those exercise so the companion from Aether is very good.

Paul also described how Aether engaged residents with limited mobility or speech by moving around, playing karaoke, or dancing. Paul shared his experiences with two residents, Eileen and Benjamin, as examples.

“So even for clients who cannot speak, they are visually engaged by Aether. Because Aether shows movements-that is very useful too. They do not communicate back, but they can still see there is an object in front of them, and interacting with them… For Eileen, she finds Aether funny every time there is music and dance comes out of it. She interacts throughout the time. When we are singing and you have Aether play Karaoke, Eileen was engaged with her verbal prompts, also clapping hands and laughing. She may not know what is happening, but she does understand it is a happy moment. Same for Benjamin. Even though he did not verbally respond to Aether, I did observe a few times that when we were dancing and singing, he was pretty engaged in it. He was focused on what was going on, and looking in the direction of Aether. He is engaged with Aether’s movements.”

3.1.3. Support Staff’s Care Delivery

Staff shared how Aether supported the care team by interacting with and engaging residents. For example, Care Aid Christy mentioned the example where Aether motivated Alice to perform exercises.

“It is nice to have Aether for a home like this. When everyone is busy, we have somebody like Aether to interact with the client, it is nice. It is a big help. It makes difference. Especially for Alice. Before Aether was here, we need to tell her ‘You have to do your rounds, you have to.’ We will like we are bossing her around. But when Aether was there cheering for her, it is different. Now all of sudden she can be excited to do her exercise because Aether encouraged her to do so.”

Care Aid Paul and Manager Yulia shared insights on how Aether helped the care team at work by accompanying the resident, especially when the care team had limited capacity to meet all residents’ individual needs at the same time.

“As you can see, our house is different. Since all our residents are in wheelchairs, we cannot carry them all at the same time because our van only takes two at a time, and each day. They get frustrated if they have nothing to do. It is good for them when they have Aether to interact with, and entertain them. Aether is interactive, which is nice, so that residents who cannot go out that day can play or sing with it.”—Manager Yulia

“Sometimes we are busy with things like doctor’s appointments, so they [residents] cannot come out and only stay at home. The companionship from Aether can sometimes help them be away from their anxieties. For example, Alice wants to go out every day, but we cannot take any client out all the time. So when Alice is at home, she will be frustrated. But Aether is a very good distraction from her anxiety of not being able to go out. With Aether, she can be engaged with something useful, and feels that she is contributing.”—Care Aid Paul

Christy gave an example of how Aether made a difference in helping a resident, Frank, speak out for himself compared with his usual style of communication with staff.

“I think Aether has a positive impact on Frank. He actually does not like to socialize. He always has a particular style with different people, and he has preferred ones. If you are not among what he wants he will never tell you that ‘My leg is in pain’ or ‘I feel this’. But I think with Aether, he is at a better position to express himself, his feelings, and what is going on in his system. Aether is in a better position for his conversation to carry on fluently.”

Similarly, Paul mentioned how Aether helped him to learn and understand each resident’s needs and preferences through their interaction with Aether together.

“I know each person better with Aether in this learning process. With Aether being here more often, now we know how to communicate with Alice, and how to integrate into Benjamin’s personal space. You learn how they interact with things because everyone is different and we’d better find that sweet spot. We cannot commit full-time one-to-one caregiving, but with Aether here, it is almost like having an extra person on board.”

3.2. Facilitator 2: Supportive Environment and Available Resources Increases Robot Usage

3.2.1. Manager’s Buy-In and a Self-Motivated Staff Champion

During our five-month deployment, Yulia started serving as the manager of the care site in the third month of our study. She was open to having the research team and industrial partner deploy Aether in the care home and collaborate in the ongoing testing of various functions of the robot. Further, she allowed Paul to be the staff champion in the project, being responsible for coordination with the research team and leading in working with the research team in the examination of the robot. Yulia said,

“There is a log of adjustment, for example letting Aether knowing Paul and Alice. I also have Aether recorded my voice and I did some dancing with Aether.”

She further acknowledged,

“It was really Paul, the key worker, who did all those things, for example, introduction to the robot, training with [author, blinded for review] and the team, everything.”

Care Aid Paul was also capable of successfully having different staff members start using Aether with residents regardless of staff rotation:

“We do not know who is working on which day… It is always alternating. But I can have my colleagues experience Aether firsthand, just getting them out there and say ‘Hey, can you go assist this client today with Aether?’ I find that is helpful because even they are coming on a different day, they would just still have that memory of what they did with Aether previously, and they will have a client interact with Aether. It is a smooth transition.”

3.2.2. Staff’s Positive View on Aether’s Role in Care Delivery

Notably, staff members were not too concerned about the possibility of Aether replacing them, as they understood the complexity of their care delivery to residents in the care setting and their responsibility. For example, Susan said,

“In this kind of setting, in this care home, it is not too easy to have Aether completely replace us. Especially in our protocol-we are one-to-one to our clients. It is also our liability to keep track on clients’ health. We can not just reply on the robot in this setting. Because we know the ability what our clients can do and what the robot can do.”

When imagining Aether’s development and upgrading in functions in the future, Care Aid Paul viewed Aether to be supplementary to the care team in addressing the psychosocial needs of residents:

“For us, we are staff, but usually we go beyond our routine to help clients. We take a step further to help them and attend to every need that they have. So, Aether could support us by giving emotional support to residents in the future. Not to a point that Aether totally replaces a human, but to a point that helps in this care setting.”

3.3. Facilitator 3: Appropriate Engagement and Training Strengthens Partnership

3.3.1. Consistent In-Person Supports

As mentioned in the methods, between February and June 2024, the research team conducted biweekly in-person visits to the care home and tested Aether’s various functions. When Aether was performing its tasks for residents, for example, playing karaoke or exercising, the research team, industrial partner, and staff joined the residents to enjoy the interaction with Aether together (e.g., singing or moving together). Manager Yulia and Community Support Worker Sharon shared their insights about the impact of such in-person interactions on improving companionship to residents from both the robot and the research team:

Yulia: For Alice, we always encourage her to do an extra exercise. In the morning, even she knows that it is her routine, she will get bored. But with you and your team cheering her and Aether following her, this part makes a difference. And plus the karaoke part that she loves. The time of singing together.

Sharon: Everybody needs attention. For Frank, when all of you come, and when Aether comes, it is kind of extra person come to him. He feels people keeps attention on him. That makes him and same for other residents feel it is a comfortable companion.

Similarly, Care Aid Paul said,

“While you are here, doing activities with Alice, it just enhances her experience. It makes difference. This way is good for our residents to have a companion, to have people around them all the time.”

In addition to in-person support, in the first three weeks of February, the research team and industrial partner also remotely supported staff to operate Aether for residents, which means Aether was tested three times a week during that period. However, staff preferred in-person support over remote ones. For example, Paul mentioned,

“When you are here and we do it together, it is easier for us. You always have a lot of students here assisting throughout the process, so of I have any questions, there is always someone here and we can always ask you. I like it when the support is reachable. But when you are not here and we do it over the phone, it takes both you and my side more time to describe and figure out which is which. Time is a big thing for me as we are all busy taking care of my clients.”

Further, he described his experiences of being consistently engaged in the project and learning new knowledge, which helped him and the care team in care delivery.

“In the beginning, when Aether was just here, there was so many updates need to be done to adapt to this house, especially to adapt to our clients. And you and your team have been testing Aether with our clients and invited me to join. But now I cannot say we know the whole thing about Aether (laugh), but we know some knowledge, about how to interact with Aether together with clients. I find that is very helpful. It is good for clients to have a companion throughout the day.”

3.3.2. Supporting Materials and Acknowledgement

Staff agreed that supporting materials such as stickers and guidelines designed by the research team provided them with clear instructions to operate and interact with Aether (see Figure 3). To ensure staff acceptance of these materials, the research team brought the draft to the staff and asked for their feedback during the visit, amended it accordingly, and showed it to staff again in the next visit.

Care Aid Paul mentioned how these materials help him to use the robot:

“Before having those things [materials] from you, I do not always remember the exact word to activate Aether. Now with what you brought to us, it is easier. You guideline shows the operation is pretty straightforward and I just follow the steps there. Yeah, these materials help and thank you for doing this.”

Care Aid Christy described how she liked the gifts from the research team to her and other staff members as a token of appreciation:

“I love the gift you gave to us. My daughter also graduated from [University name blinded for review] and one time, when I got the hoodie from you, I showed to her. She asked ‘Mom, why do you have that?’ I said ‘We have a robot that comes to the house and the students gave this to us.’ Also thank you for the bubble tea brought to us. It is my favourite! (laugh)”

Notably, a strong partnership between the care home and research team facilitated by strategic engagement also enables the research team to identify more residents’ needs from staff. For example, as of May 2024, we had Aether perform various tasks for our residents, including singing karaoke, dancing, telling jokes and riddles, and making up stories and poems together. In one of the site visits that month, we asked Community Support Worker Sharon for additional functions she wished Aether would have to better support residents. Sharon mentioned,

“It will be great if Aether can motivate Alice to exercise more to train her arm strengths -I mean moving in loops in this home. Because currently we find it hard to motivate her, probably because we see each other everyone. For us, it is also tiring for us to keep encouraging ‘Good job Alice! Now can you come to me, to here?’ But with Aether, I think maybe it can help. If Aether can, for example, move in the front and invite Alice to follow, or stand behind her to motivate her moving forward, it will be a new experience to her and she can have fun while exercising, even Aether says the same thing that we say.”

Based on Sharon’s request, the research team and industrial partner added the exercise function in Aether and tested it with staff and resident Alice. In our visits, we observed that with Aether, Alice completed between 6 and 7 loops on average, which is higher than her record of 2 to 3 loops before Aether was introduced, according to staff.

3.4. Barrier 1: Non-Customized Design Hinders Aether Meeting Diverse Needs

3.4.1. Limited Proactivity and Support to Mental Health

Currently, the robot is designed to be reactive, making interaction with Aether conditional and heavily relying on staff to start. Given the context and residents’ demography of the care home, staff mentioned that the robot needs to be more proactive and initiative to better serve residents and support the care team. For example, Community Support Worker Sharon wished Aether could be programmed to initiate a conversation session for residents.

“For me, one of the barrier is that you need to command Aether before its service being carried out. If you do not command, nothing happens. For our residents, they will not go left if the staff does not tell them to move left. If you are not beside them, they will not use Aether. So staff has to be there to command. Maybe in the future, Aether can do something for our residents in our absence. If say 3 p.m. every day is we know when all residents are in this home, Aether can start conversation with them and they can come around and like having a meeting with Aether. That would be better.”

Similarly, Care Aid Paul hoped Aether was more proactive:

“In the future, I hope it (Aether) will be unprompted, like just happens automatically, instead of we prompting Aether to do something. It can recognize Alice, like, ‘How is your day Alice?’, or when Alice passes by and Aether can start a conversation with her. More intelligent (laugh).”

Social Worker Susan further imagined that more proactive Aether can be better integrated into the routine of residents’ daily lives or staff’s work and contribute to both parties.

“The care home can put the robot as a daily practice, and I think that is the best approach to get the clients familiar with the robot. I think of one example from another (care) home. I work regularly with a client, who has a routine to brush her teeth everyday at 4 p.m., but I always forgot to give her the toothbrush. So just an example, if some programming can be done on the robot, who can move around the house and remind a client to brush her teeth. It can also ask the client to ask staff to give the toothbrush and toothpaste, something like this. Reminder is something good-it is also a kind of interaction. Also, if a staff is organizing like a music group, and the robot can join as a participant in such activities, that will make the whole event more interesting as well.”

In addition, staff expressed their desire for improved mental health support from Aether to residents. For example, Sharon imagined a scenario where Aether autonomously addresses the mental health challenge of a resident, Frank, who faces challenges voicing his feelings.

“Maybe if Aether is beside Frank, and be able to identify that Frank is crying and speak out ‘How are you doing? Are you ok Frank?’ That will be better.”

Paul hoped Aether could support the mental health of residents with flexible approaches:

“I hope Aether can understand the client more and what type of moment they are having. For example, Frank is crying, and maybe Aether can recognize ‘He is having a bad moment’, and can just play some type of music that would calm him down. So I hope Aether could have some sense of certain type of emotion, for example, what crying is. Even when it walks around, like, sees a sad face on Alice, it could ask questions like ‘Hey Alice, are you doing ok?’ ‘What can I help you throughout the day?’ ‘Should I tell you a story?’ So Aether does not have to physically do anything, and it could just be there emotionally.”

3.4.2. More Adaptive and Inclusive Interaction Is Needed

Managers and staff members pointed out that Aether needs to be more flexible and accommodating in conversation with care staff and residents in the care home.

“Sometimes Aether does not understand Alice when she speaks low. Also, it would be easier and beneficial if Aether can accommodate different accents. You may have noticed we are multicultural. Each of us has an accent, but Aether cannot distinguish that. Also, it is better if Aether can read hand signs like a wave, so those who are not verbal can also start interaction with it.”—Manager Yulia

Social Worker Susan explained that the conversation function in Aether needs to be more adapted to the residents and the settings in the care home:

“I think the robot is designed for care homes where clients are more independent-who can walk closely to the robot and talk, go upstairs and downstairs, and able to speak verbally fluently. Last week, I remembered when Aether asked Alice ‘What is your name?’ And Alice keeps saying ‘I am Alice, Alice’. And the robot asked again, ‘Is it Alice?’ And then Alice said ‘Yes I am.’ I think the robot cannot identify exactly the way she speaks. Also, maybe because our residents are all on wheelchair, they are not possible to speak closely to the robot, and one of you need to repeat closely to the robot for Alice.”

Care Aid Paul also elaborated,

“Even there is a complicated plan, a big plan, as long as it is communicated in simple English, it is fine. And I hope its language can be tailored to each client. For example, Alice would not understand if you just go on with a long sentence. She would not. Aether needs to talk to her in simple English, the way she understands. It needs to go in a step-by-step process, not put everything on her.”

Additionally, Care Aid Christy was concerned that the conversation-based interaction programmed in Aether was impacting equity among residents with different levels of impairments in receiving stimulation and benefiting from Aether. For example, Christy worried that Aether is less available to other residents compared with Alice, who is the only resident capable of speaking in the care home.

“Currently, the smile that Aether has is very good. It is a big thing for them. But I hope It will be nice if Aether would go from one person to another, even just 5 min for a ‘Aether time’, for example, singing together, or play music in front of them. Because Alice can ask ‘Hi Aether, can you sing with me?’ But for Frank, Eileen and Benjamin, they cannot do that. If I do not talk to them, they will just sit there and will just fall asleep all day. They are the ones that need us to find a way to stimulate them. It will be nice if they are also part of the music therapy by Aether. That just 5 min of music, or showing something that will stimulate them. They will probably go crazy.”

Manager Yulia and Community Support Worker Sharon expressed their expectation for more sensory stimulation in Aether, such as tactile and audio ones:

Sharon: Frank likes to shake hands and touch fluffy stuff.

Yulia: Yes, I hope Aether have features like hands and allow people to touch. I hope it feels like petting a cat or a dog, comfortable and warm.

Sharon: Yeah! Like woof woof or meow.

Yulia: That would be good. These are basic things. It can even talk to clients things like: here is sky. It is blue. Here are roses. They are red.

Sharon: Yes, I told industrial partner that currently, the voice of Aether is robotic. I hope he can change this.

3.4.3. Not Tailored to the Context or Helpful to Staff’s Work

Staff think that Aether in the current stage is not customized for the needs of residents and staff in their daily lives in the context of the care home. Community Support Worker Sharon explained why the fall detection function in Aether is not tailored to this care home:

“This (fall detection function) is for residents with high functioning, walk by themselves and go to washrooms by themselves in other care homes. These mobile residents may fall and Aether can tell the staff. Here our residents are all in wheelchairs, so I think right now we do not need it (fall detection function) in Aether.”

Moreover, Sharon would have preferred if Aether could detect security hazards around the care home: “ I hope Aether is able to identify, for example, a dangerous person enters or walk around the house.”

Additionally, staff wished Aether could help them to know their clients better, as Care Aid Paul said,

“I hope it helps us know more about our clients. I hope I can just ask Aether questions like, ‘’Can you let me know what Alice needs, and what she loves?”

3.5. Barrier 2: Insufficient Structural Supports and Resources Held up Further Usage

Despite the aforementioned strengths and enablers in the robot design and care home environment, the care home faces several challenges in deploying Aether.

3.5.1. Limited Human and Infrastructural Resources

The care home received limited support in human resources. Besides the industrial partner and engineering students from the research team, there were no external experts in IT or engineering assigned to support the care home in deploying Aether during our 5-month study at the care site. The majority of the troubleshooting and maintenance of the robot was performed by the research team in partnership with the industrial partner. Also, during our study period, there was little structural support, such as training and educational opportunities, to improve the care team’s digital literacy and prepare managers and staff members for the deployment of the robot. Further, the care team faced high staff turnover, which is discussed in later sections.

In terms of infrastructure, a strong and stable internet connection is needed to support Aether in performing functions such as AI-driven interaction with users and detecting safety hazards in the environment. In our study, the high-quality internet was set up and maintained by the industrial partner.

3.5.2. Unclear Responsibility and Teamwork

When speaking about Aether’s detecting safety hazards, Community Support Worker Sharon expressed her confusion on unclear responsibility among staff brought by Aether:

“Let’s say Aether detected something dangerous, but not all staff will be able to follow up on it. How will others recognize or view the danger? Can we leave it ‘unsafe’? Of course, if Aether can fix the danger, I will be really happy for that.”

3.6. Barrier 3: Staffing Crisis Interrupted Deployment and Exploration

During our study, we experienced high managers and staff turnover in the care home, which increased the input of time to introduce Aether to new staff members and motivate them to use it. Though the research team held a few training sessions for staff, the instability in staffing made it difficult for every staff to be prepared to use Aether.

Care Aid Paul said,

“We have different staff come in each shift, some of them are even casuals. It will be easier for staff to look at our clients interact with Aether, so they know how to communicate with Aether and Alice, or Aether and Benjamin. It will be. It is better for everybody to be on board at the same time, but I also know that will be hard.”

Community Support Worker Sharon also mentioned how high turnover made it hard for all staff to receive training on the operation of Aether:

“In here, most of the staff come and go. That becomes a barrier for most of staff not having an idea of Aether. If that an opportunity of staff training session is given to all the staff to having a conversation with Aether, or knowing how to operate Aether, I am sure it will be better.”

4. Discussion

This study investigates the feasibility and acceptability of deploying an AI-driven collaborative service robot, Aether, in a group home for residents through the lens of staff. We identified three key facilitators made deploying Aether is feasible in the care home: (1) Helpful Robot Features Support Care Delivery, (2) Supportive Environment and Available Resources Increases Robot Usage, and (3) Appropriate Engagement and Training Strengthens Partnership. Our results challenged assumptions about residents’ acceptance of an AI-enabled robot in LTC. Further, residents and staff accept Aether in their daily routine and can potentially benefit from the innovation. We also identified several barriers to use Aether in the care home, and recommended future directions for research and practice.

4.1. Innovation Features

It is imperative that AI-enabled robot is designed to align with the complex, contextual needs of the care environment. In our study, the most frequently used functions in Aether by residents and staff were for engagement, companionship, and supporting residents’ mental health and overall well-being, despite Aether’s original design focusing on technical tasks (e.g., detecting safety hazards). This shift in usage underscores the priority of addressing residents’ psychosocial needs within AI-enabled innovations and highlights the importance of equipping staff with innovation and resources that can support these needs. Our findings resonate with existing research that emphasizes the critical need for social connection among residents in LTC homes, reinforcing the importance of designing AI solutions that could foster such connections and support staff’s care practice [25,35].

Another advantage of Aether is its strong adaptability driven by AI, which benefits innovation recipients (staff and residents) in two ways. Firstly, it enables staff to utilize its various functions to offer individualized experiences to residents with different levels of capacity and preferences, and to know more about each of them. For instance, Aether’s storytelling and karaoke features can engage Alice, who is capable of communicating verbally, by making up stories and singing together, and she becomes more confident to speak out her needs through this process. Other residents with speech impairments can also be engaged by Aether when it is singing, dancing, and moving around, according to staff. Moreover, adaptability means an improved familiarity with the innovation to users as their interaction with the innovation increases. We observed how Aether was accustomed to users’ tones, and their conversation with Aether became smoother as the deployment proceeded. As a result, users were encouraged to interact with Aether more.

Despite its advantages, AI-enabled innovation needs to be more compatible with the care home (inner settings). One key area for improvement is fostering inclusivity in user interactions to promote digital equity, particularly for users with diverse capabilities. In our project, interaction with Aether is mainly verbal. Interaction based on certain capabilities (speaking) may disadvantage users who are less capable in that function (e.g., those with speech impairments). Our findings suggest that to improve accessibility and promote digital equity, AI-enabled innovations should allow residents to initiate interactions in a variety of ways—for instance, through simple gestures, a single word, or by pressing a button. The staff also learned more about their residents through interacting with Aether. This may create an opportunity for staff to care for LTC residents innovatively. AI-enabled innovation can be more proactive in offering services to residents (e.g., playing familiar music) instead of passively waiting for commands. Our findings corroborate with studies on the LOVOT robot, an AI-enabled robot that does not talk [19]. Researchers identified that LOVOT can provide users comfort and companionship, reduce loneliness and anxiety in LTC settings, and promote emotional well-being through touch, movement, and other sensory interactions [19]. Secondly, Aether currently has difficulty recognizing English spoken with accents, which could hinder its accessibility for users who speak with non-standard accents or who use English as a second language. Given Canada’s multicultural population, especially a diverse workforce in LTC homes, it is crucial that AI innovations like Aether are designed to accommodate a wide range of accents and languages [36]. Future iterations could prioritize more inclusive language recognition, ensuring that staff and residents from various linguistic backgrounds can interact with AI systems more seamlessly. Also, it may be interesting to investigate the integration of deep learning technology for speech identification for service robots in group homes, as this may address the reported lack of adaptability and promote digital equity, allowing all users to fully benefit from the technology [37,38].

4.2. Partnership and Relational Connection

Our study findings on facilitators and barriers also suggested the significance of partnership and teamwork between researchers, care homes, and industrial partners. In terms of facilitators, all parties shared a common goal: to continue using Aether to maximize its potential benefits for residents and staff. Each group made concerted efforts to build strong relational connections to move towards the shared goal. For example, the care home manager played a pivotal role by actively supporting the regular deployment and testing of Aether, while also designating a motivated implementation lead—identified by the research team—to coordinate with researchers, test Aether, and influence other staff members to embrace the robot.

Before deployment, the industrial partner equipped Aether with interactive AI-driven features (e.g., everyday conversation), which allowed staff and residents to immediately see the robot’s value and helped sustain momentum for its continued use. The research team, following rigorous research protocols, tested Aether, and identified the specific needs in the care setting with complexity. We employed strategies to engage the care team with respect, recognizing their contributions and providing them with regular in-person support and relevant materials to facilitate their use of the robot. These strategies were appreciated by the staff, who viewed them as essential in enabling the ongoing deployment of Aether. The combined efforts from each party contributed to building a sense of “teamwork” or “community” and led to positive deployment outcomes [39]. Our findings also corroborate the existing literature on strategies to establish and strengthen partnerships to successfully deploy innovations and shed light on the key facilitators and barriers in the LTC setting [40,41,42].

Moreover, strong partnerships and relational connections can temporarily overcome barriers in our study. For example, there were not sufficient human and infrastructural resources during the deployment of Aether, reflected by high staff turnover, lack of IT experts and training opportunities, and weak internet connection. The research team delivered training sessions and provided staff-friendly guides, and sometimes the care team allowed researchers to have Aether interact with residents when staff were not available. In the future, researchers are recommended to investigate the impact of robot integration on workplace dynamics in different and care sites and care environments. Further, policymakers are recommended to work with care home leaders, staff, industrial partners, and residents in developing strategies that nurtures collaboration and partnership in deploying different technology, including AI-enabled robots, in LTC homes.

4.3. Support and Resources from Inner and Outer Settings

Our findings implied some barriers hinder the deployment of Aether, especially sustainable deployment due to insufficient structural support and resources. The results from this study could help inform and expand future policies and practice guidelines when implementing service robots in different healthcare settings. Our study results could guide the policies and practice guidelines on maximizing identified facilitators while reducing identified barriers. For example, in terms of human resources, lack of sufficient care staff, external IT experts, trainers, and educators in nursing practice are threats to sustainable deployment of innovations. Without enough frontline staff members to provide sufficient care to residents, the care team may face challenges such as overloaded work and high staff turnover [7,43,44]. In that case, using AI-enabled innovation becomes less prioritized compared with other tasks, which may negatively impact deployment outcomes [45]. Due to the machine learning nature of AI, the less staff interact with the innovation, the less AI knows about its users and the more difficult for it to tailor to the needs of residents, which further discourages staff from using it. In the long term, this may threaten digital equity for users, including those in LTC homes, to benefit from AI technology [46]. Our results also suggested another impact of using AI is that staff are confused by suggestions and recommendations by AI. This implies that LTC homes, and potentially other institutional care settings, need to be prepared for changes in hierarchy, training, and power dynamics brought by AI. Further, by recognizing the importance and benefits of service robot implementation in healthcare, necessary education initiation from school in different fields (including medicine, nursing, and public health) may help prepare future healthcare providers and care leaders to become more familiar with AI-enabled robots and build competency in using such robots to facilitate patient–provider communications and expedite the work. Training from the institutions and facilities will also help increase the knowledge levels of service robots in the current workforce and will synergize with already educated incoming healthcare providers [47]. We call for more structural supports to care homes from multiple partners (e.g., health authority leaders and policymakers) in terms of human, technological, and infrastructural resources to prepare for the adoption of AI from the macro level [37,48].

5. Strengths and Limitations

5.1. Strengths

This project advanced the knowledge of feasibility and acceptability of deploying an AI-driven service robot from staff’s perspectives for residents in a group home. The participating staff shed light on the often-overlooked complexities in meeting residents’ unique needs and staff’s needs in care practice. Further, we deployed Aether for 5 months, gaining valuable insights into the factors that influence its integration into the care home, as it interacted with different residents, activities, and scenarios. Our study participants included care staff and manager from different backgrounds to gain a comprehensive understanding of facilitators and barriers to deploying AI-driven innovation. During our deployments, we involved the residents, staff, and manager from the care home, industry partners, multidisciplinary researchers, and older adult partners. This collaborative approach ensured that meeting residents’ needs and supporting staff in care delivery are prioritized, ensuring the innovation is modified to be better adapted to the care home’s routine and context. Our study contributes to further practices and research of deploying and implementing AI-driven innovation to traditionally disadvantaged populations (e.g., residents with disabilities or impairments in institutional care sites) to increase their benefit from AI technology, improve their well-being, and promote digital equity.

5.2. Limitations

This study is subject to the following limitations. Some staff members may be reluctant to share barriers for them to use Aether in their own workplace. Fear of judgment or repercussions may have led to the underreporting of negative experiences. There are potential power dynamics in the paired interview in which the Manager was interviewed together with the Community Support Worker. Another limitation is selection biases. Staff with more interaction with Aether were more likely to have good experiences with Aether and more likely to participate in interviews, which may influence results, especially the part about positive experiences with Aether. Further, the study required staff who were proficient in English due to the nature of Aether’s interactions. This may have limited the sharing of valuable insights from staff with limited proficiency in English. Moreover, the study was conducted in a small group home in an urban area. Findings from the study may not be generalizable to broader populations due to the specific context, demographic characteristics, and small sample size involved in the research. Our detailed description of the study context enables readers to assess the applicability of findings and make informed judgments about the relevance of the insights in relation to their unique circumstances. Lastly, CAR may have some limitations. For example, in our context, CAR requires intense and ongoing training on multidisciplinary trainees to work with different partners. This includes our industrial partners who developed Aether, LTC staff, and residents-a vulnerable population with comorbidity and diverse needs. The resources and time required in CAR may not be generalizable to other research projects and contexts.

6. Conclusions

Our study aimed to investigate the feasibility and acceptance of deploying an AI-enabled service robot in an LTC home from staff perspective. Facilitators and barriers shaping the feasibility were identified. While Aether provided helpful features to residents and staff, such as engagement and accompany residents, and support staff’s care delivery, the pressing barriers, for example, limited resources and supports available to the care team, need to be addressed. This study provides a positive starting point for the service robot implementation in a LTC home. Future research can further explore the deployment of AI-enabled robot for different populations, especially vulnerable populations and at high risks of being marginalized in accessing and benefiting from AI technology. Also, there is a need to further investigate the voice of staff and managers in LTC to inform further integration of AI-enabled service robots in care delivery. Emerging initiatives are recommended to address barriers faced by staff and care sites in deploying AI-enabled service robots. We call for more cross-sectional collaboration to prepare care homes for an era where AI-enabled service robots are expected to make bigger contributions in improving the well-being of LTC residents.