Human-AI Co-Ideation Across Socio-Technical Configurations: An Interactional Analysis of Planar and Immersive Workspaces
Organizations: Computer Science, Wellesley College, 106 Central st., Wellesley, 02481, MA, USA · Business School, Harvard, Boston, MA, USA · Electrical and Computer Engineering, University of New Hampshire, Durham, NH, USA · Information Systems, University of Haifa, Haifa, Israel
Abstract
Remote collaborative platforms increasingly integrate generative AI to support group ideation. These platforms are sociotechnical configurations in which spatial mediation, interaction modality, AI output format, and the mechanisms that make contributions visible to a group co-vary as properties of a deployed system. This paper examines how such configurations shape participants' collaboration and their engagement with AI during creative work. We report a comparative study of two commercially available platforms that instantiate contrasting configurations: a planar shared canvas with text-based AI integration, and an immersive virtual reality workspace with concise AI output embedded in the shared space. Fourteen all-women triads completed a structured ideation task consisting of human-only idea generation, AI-supported idea expansion, and group convergence. We analyzed collaboration patterns, selection of AI-generated ideas, perceived idea quality, workload, presence, and their lexical characteristics. The immersive configuration yielded greater spatial presence and introduced a participation paradox, in which spatial freedom enabled exploration and fragmented group coherence; it was also associated with more balanced role distribution during convergence, at the cost of higher physical demand. In both configurations, AI multiplied the idea pool, and ideas groups authored with AI available were rated above participants' own. AI output format: about 21 words per idea on the planar canvas and 3 in the immersive workspace. Tracked how participants evaluated AI-generated ideas and how often they selected them unaltered as finalists. Neither configuration made AI contributions visible or attributable to the group: participants invoked AI individually and asynchronously in both. We derive design implications for AI output format, group-visible AI invocation, and spatial support for convergence
Figures & tables
| Term | Denotes |
|---|---|
| group, triad | the three human participants in a session |
| collaboration, team | relations among the human participants |
| co-ideation | the joint activity of a group and the AI system during idea generation |
| co-elaboration | the act of a participant extending, combining, or generating new ideas from AI output |
| Source | Origin & Timing | Curation Workflow | n |
|---|---|---|---|
| Human | Entered by individual participants during Subtask 1 (human-only divergence); snapshot frozen at the end of Subtask 1. | Deduplicated identical notes; prompt-label nodes removed; word count recorded. | 308 |
| AI | Inserted by the platform’s AI feature during Subtask 2 (AI-supported divergence) when a participant invoked it (”Expand with ideas” in Miro; the ”brainstorm” node in Noda); snapshot frozen at the end of Subtask 2. | Formatting artifacts stripped (for example, Miro’s leading asterisks); word count recorded. | 937 |
| Group-authored | Distinct from both Human and AI categories. Authored or edited by group members during Subtask 2 after reviewing the pooled human and AI ideas; snapshot frozen at the end of Subtask 2. May draw on AI output, but contains no unmodified AI text. | Word-count recorded. | 15 |
| Platform | Human Ideas | AI Ideas | Group-authored Ideas | Total |
|---|---|---|---|---|
| Miro | 153 | 304 | 7 | 464 |
| Noda | 155 | 633 | 8 | 796 |
| Type | Miro: Mean (SD) | Noda: Mean (SD) | Cohen’s (Noda vs Miro) |
|---|---|---|---|
| AI ideas | 21.41 (6.37) | 2.88 (1.70) | |
| Human ideas | 5.03 (5.16) | 3.50 (2.08) |
| Quality Dimension | Idea Source | Mean | SD | F | df | p | Pairwise Comparisons | |
|---|---|---|---|---|---|---|---|---|
| \multirow 3* Relevance | Human | 4.33 | 0.73 | \multirow 3*2.30 | \multirow 3*2,40 | \multirow 3*.113 | \multirow 3*0.103 | \multirow 3*No significant differences |
| AI | 4.24 | 0.77 | ||||||
| Group-authored (Team) | 4.62 | 0.50 | ||||||
| \multirow 3* innovation | Human | 3.38 | 1.02 | \multirow 3*4.02 | \multirow 3*2,40 | \multirow 3*.026* | \multirow 3*0.167 | \multirow 3* Group-authored AI (p .05); |
| AI | 3.38 | 1.20 | ||||||
| Group-authored (Team) | 3.95 | 0.86 |
| Quality Dimension | Idea Source | Mean | SD | F | df | p | Pairwise Comparisons | |
|---|---|---|---|---|---|---|---|---|
| \multirow 3* Relevance | Human | 4.62 | 0.50 | \multirow 3*15.08 | \multirow 3*2,40 | \multirow 3* .001*** | \multirow 3*0.430 | \multirow 3* Human AI (p .05); Group-authored Human (p .05); Group-authored AI (p .001) |
| AI | 3.95 | 0.92 | ||||||
| Group-authored (Team) | 4.95 | 0.22 | ||||||
| \multirow 3* innovation | Human | 3.52 | 0.93 | \multirow 3*23.00 | \multirow 3*2,40 | \multirow 3* .001*** | \multirow 3*0.535 | \multirow 3* Human AI (p .05); Group-authored Human (p .01); Group-authored AI (p .001) |
| AI | 2.62 | 1.24 | ||||||
| Group-authored (Team) | 4.29 | 0.56 |
| Idea Quality Measure | Platform | N | Mean | SD | t | df | p | 95% CI | Cohen’s d |
|---|---|---|---|---|---|---|---|---|---|
| Human-generated ideas | |||||||||
| Relevance | Miro | 21 | 4.33 | 0.73 | -1.48 | 40 | 0.146 | [-0.68, 0.10] | -0.46 |
| Noda | 21 | 4.62 | 0.50 | ||||||
| innovation | Miro | 21 | 3.38 | 1.02 | -0.47 | 40 | 0.638 | [-0.75, 0.47] | -0.15 |
| Noda | 21 | 3.52 | 0.93 | ||||||
| Insightfulness | Miro | 21 | 3.43 | 0.98 | -0.93 | 40 | 0.357 | [-0.91, 0.33] | -0.29 |
| Term | AI ideas | Human ideas | Test | Statistic | Sig. |
|---|---|---|---|---|---|
| promote | 68 | 0 | Fisher | ||
| practice | 84 | 2 | Fisher | ||
| organ | 64 | 1 | Fisher | ||
| host | 29 | 0 | Fisher | ||
| sus | 0 | 6 | Fisher | ||
| hall | 19 | 16 | chi-square | ** |
| Term | Noda ideas | Miro ideas | Test | Statistic | Sig. |
|---|---|---|---|---|---|
| economy | 11 | 0 | Fisher | ||
| organ | 3 | 62 | Fisher | ||
| encourage | 2 | 46 | Fisher | ||
| install | 0 | 28 | Fisher | ||
| focus | 0 | 21 | Fisher | ||
| college | 19 | 3 | Fisher |
| AI | Human | Group-authored | ||
|---|---|---|---|---|
| \multirow 3*Miro | High Cost | 3 | 1 | 3 |
| Medium Cost | 3 | 2 | 2 | |
| Low Cost | 4 | 1 | 2 | |
| Miro Total | 10 | 4 | 7 | |
| \multirow 3*Noda | High Cost | 0 | 2 | 5 |
| Medium Cost | 3 | 2 | 1 | |
| Miro | Noda | |||||||
|---|---|---|---|---|---|---|---|---|
| Item | Miro | Noda | Mdn | IQR | Mdn | IQR | ||
| In the computer generated world I had a sense of “being there”. | 20 | 21 | 4 | 2–4 | 6 | 5–6 | 56.5 | .001*** |
| How real did the virtual world seem to you? (first occurrence) | 19 | 21 | 4 | 4–5 | 4 | 3–5 | 235.0 | .325 |
| I did not feel present in the virtual space. | 19 | 21 | 4 | 3–4 | 5 | 3–6 | 140.0 | .102 |
| How aware were you of the real world surrounding while navigating in the virtual world? (i.e. sounds, room temperature, other people, etc.)? | 19 | 21 | 4 | 2.5–4 | 5 | 4–6 | 113.0 | .017* |
| How much did your experience in the virtual environment seem consistent with your real world experience? | 19 | 21 | 4 | 3–4 | 3 | 2–4 | 233.5 | .323 |
| Miro | Noda | |||||
|---|---|---|---|---|---|---|
| TLX item | Mdn | IQR | Mdn | IQR | ||
| Mental demand: How mentally demanding was the task? | 4.00 | 3.10–4.90 | 4.30 | 3.40–5.20 | 180.0 | .312 |
| Physical demand: How physically demanding was the task? | 1.00 | 1.00–1.90 | 1.90 | 1.30–3.70 | 126.0 | .015* |
| Pace: How hurried or rushed was the pace of the task? | 4.30 | 3.70–4.90 | 4.30 | 3.10–4.90 | 241.5 | .603 |
| Success: How successful were you in accomplishing what you were asked to do? | 5.20 | 4.60–5.50 | 5.20 | 4.30–6.10 | 212.0 | .840 |
| Effort: How hard did you have to work to accomplish your level of performance? | 4.30 | 3.40–4.90 | 4.60 | 3.10–5.20 | 206.5 | .733 |
| Miro | Noda | |||||
|---|---|---|---|---|---|---|
| Item | Mdn | IQR | Mdn | IQR | ||
| To what extent did you feel that team decisions were made collaboratively, taking into account different perspectives? | 4 | 3–4 | 4 | 4–5 | 160.5 | .108 |
| Did you feel comfortable providing feedback to your team members and receiving feedback from them? | 4 | 4–5 | 5 | 4–5 | 148.0 | .047* |
| How well do you think team members understood and appreciated each other’s strengths and weaknesses? | 3 | 3–4 | 4 | 3–4 | 182.0 | .322 |
| How much of the time did you feel that everyone’s ideas were encouraged and considered during team decision-making? | 4 | 4–5 | 4 | 4–5 | 235.0 | .694 |
| On a scale of 1-5, how satisfied are you with the overall level of collaboration within the team? | 4 | 4–5 | 5 | 4–5 | 152.0 | .063 |
| Miro | Noda | |||||
| Item | Mdn | IQR | Mdn | IQR | ||
| Own ideas | ||||||
| The ideas I contributed were relevant to the problem. | 4 | 4–5 | 5 | 4–5 | 174.5 | .190 |
| The ideas I contributed were innovative. | 4 | 3–4 | 4 | 3–4 | 203.5 | .657 |
| The ideas I contributed were insightful. | 3 | 3–4 | 4 | 3–4 | 180.5 | .295 |
| AI-generated ideas | ||||||
| Theme | Code | Participant | Quote |
| No challenges | None | P_3 (Miro) | There was pretty much no challenges. We all helped each other and listened to each other’s ideas. |
| Unfamiliarity | Unfamiliarity with teammates | P_45 (Noda) | It was harder to exchange ideas with the person I didn’t know prior to the experiment because I didn’t know their name |
| Application navigation | P_28 (Noda) | It was a bit awkward moving around the space at first, and all of our nodes were in different areas, so I couldn’t really see what other people were thinking without moving around. | |
| P_1 (Miro) | Just navigating the site, learning different short cuts | ||
| Lack of subject knowledge | P_33 (Miro) | It was a bit slow at first because we do not know each other, and we do not know a lot of the specifics of the problem we were tasked to solve | |
| Disconnection with Teammates | Perceived virtual distance | P_29 (Noda) | Not being able to physically move to see each other’s ideas; being really far away in the beginning, etc. |
| Theme | Code | Participant | Quote |
|---|---|---|---|
| Positive | Quick | P_13 (Miro) | We only had 3 minutes to generate ideas and in that time each only had 2-5 ideas, whereas using AI we each found 10+ ideas in seconds. |
| Many ideas | P_16 (Noda) | They contributed by generating many ideas. | |
| P_14 (Miro) | Using AI greatly expanded our idea pool, although sometimes it gave redundant or un-innovative answers. | ||
| Provided a starting point, simplified | P_25 (Miro) | It helped give a starting point for the individuals and let us think more creatively about ideas we would not have come up with by ourselves. | |
| P_43 (Noda) | the AI features didn’t necessarily come up with concrete ideas but they did provide me with good building blocks, like it suggested cost saving and that made me think of doing a clothing share […] | ||
| Insightful or high quality | P_20 (Miro) | I thought AI contributed to the idea generation process because it added project ideas I hadn’t considered, such as a sustainable fashion show. |
| Theme | Code | Participant | Quote |
| Dialogue | Agreement between teammates | P_20 (Miro) | I thought it was great to get to brainstorm solutions to a problem I hadn’t considered in depth before with a group of people, and I thought we came up with some great ideas together. I had a positive experience because I felt that all our perspectives were represented in the final product. |
| Compatible/respectful dialogue | P_28 (Noda) | I think we did a good job of communicating, so asking questions and offering feedback | |
| P_15 (Miro) | We didn’t talk over each other and listened to everyone’s ideas. | ||
| Task | Physically seeing each other’s ideas | P_38 (Noda) | It was fun to talk and see the physical display of ideas in front of each person, it feels like we all contributed. |
| New insight/perspective | P_42 (Noda) | I enjoyed brainstorming and learning about ideas I had not thought about. My team had prompted the AI in different ways which led to drastically different results which was interesting | |
| Ability to focus on task well | P_6 (Miro) | We were able to talk through our ideas and build on them with each other, expressing agreement or disagreement and using that to come up with a final result. |
| Theme | Code | Participant | Quote |
| Positive | Generated more ideas or expanded on previous | P_43 (Noda) | it helped each of us come up with more ideas to share between the group. |
| P_2 (Miro) | AI helped with expanding on our ideas and adding more details. | ||
| Helped break the ice | P_16 (Noda) | The AI features helped us find common ground in our ideas. | |
| Starting point for collaboration | P_40 (Noda) | It seemed we mainly used AI features beforehand that we shared with each other, but did not use while collaborating. So, I would say that AI only helped our individual selves share our ideas with each other, but it did not contribute to our full group ideas. | |
| P_33 (Miro) | The AI allowed us to easily pick from a larger set of possibilities without having to spend much time thinking beforehand. However, most of our discussion revolved around specifics which the AI did not know about. | ||
| Hindered | Required excessive pruning | P_28 (Noda) | I think that, although the AI gave a lot of ideas and suggestions, it became overwhelming how many there were, and it became a little challenging to sort through everything. |
| Theme | Code | Participant | Quote |
| No preference | None | P_47 (Noda) | Good ideas were able to come from both my team mates and the AI. |
| Preferred AI | Generated higher number of ideas | P_15 (Miro) | AI had more ideas instantly. We had a short amount of time to think of ideas, which made it difficult to fully process. |
| P_45 (Noda) | The AI generated way more answers but their answers were also way more general. So it was important to balance out its ideas with our understanding of Wellesley. | ||
| Provided a starting point | P_39 (Noda) | human generated ideas seemed more specific and particularly relevant to wellesley, but AI helped create a jumping off point for them | |
| More detailed/insightful | P_13 (Miro) | The AI generated ideas went much more in depth, whereas ours were more bullet points or non descriptive ideas. | |
| Easier to use AI | P_18 (Noda) | I feel like most of our ideas were similar to the AI generated ideas, therefore ultimately we preferred our own. |
| Theme | Code | Participant | Quote |
|---|---|---|---|
| Helpful | Applicable to future | P_3 (Miro) | I feel like I would definitely use this website for my personal planning taks in the future. |
| Good organization/visualization | P_41 (Noda) | The platform, Noda, was very helpful in brainstorming, connecting ideas, and keeping track of them. The only problem I had was that I was running out of space and it started to become cluttered. | |
| Inspired collaboration/creativity | P_14 (Miro) | The platform was very helpful in generating ideas. The mind map allowed us to branch our ideas off accordingly and the AI feature helped us generate new ideas when we were stuck. | |
| P_16 (Noda) | Noda, allowed us to visualize our thoughts, connect the, and expand on them with ease. | ||
| Starting point | P_5 (Miro) | I think Miro is a good start, but in terms of planning and giving more actions, we need more effort. not just relying on this platform | |
| P_15 (Miro) | Miro was useful because it allowed us to organize our thoughts in a way that is easier to understand. The AI tool was also helpful in this case, as I was having some trouble getting started. |
| Theme | Code | Example |
| Teamwork Strategy | Individual | M7 - Subtask 1: all participants started working on the task in separate areas of the Miro board, no collaboration in any way. |
| Equal Collaboration | M4 - Subtask 3: Participants created one central shared mind map/nodes for listing the 3 ideas. | |
| Driver/Navigator | N6 - Subtask 3: P2 verbally suggests and shares out ideas while P1 and P3 create and label the nodes. | |
| Physical state (of VR avatar or cursor) | Facing | N3 - Subtask 1: Participants made individual small maps, remained in their spawn position, but are all facing the center of the room. |
| Not Facing | N2 - Subtask 1: Participants are in a triangle formation but facing outwards. | |
| Long Distance | M5 - Subtask 1: Participants work in opposite corners of the Miro Board and their maps do not touch. |
| Key | Question | Response options |
|---|---|---|
| Q1 | How mentally demanding was the task? | Scale: {Very Low, Medium, Very High} |
| Q2 | How physically demanding was the task? | Scale: {Very Low, Medium, Very High} |
| Q3 | How hurried or rushed was the pace of the task? | Scale: {Very Low, Medium, Very High} |
| Q4 | How successful were you in accomplishing what you were asked to do? | Scale: {Very Low, Medium, Very High} |
| Q5 | How hard did you have to work to accomplish your level of performance? | Scale: {Very Low, Medium, Very High} |
| Q6 | How insecure, discouraged, irritated, stressed, and annoyed were you? | Scale: {Very Low, Medium, Very High} |
| Key | Question | Response options |
|---|---|---|
| Q19 | What challenges did you have interacting as a team? | Open text response |
| Q20 | In what ways did using AI features contribute to or hinder the idea generation process? | Open text response |
| Q21 | What positive experiences did you have interacting as a team? | Open text response |
| Q22_1 | The ideas I contributed were relevant to the problem. | {Strongly disagree}; {Somewhat disagree}; {Neither agree nor disagree}; {Somewhat agree}; {Strongly agree} |
| Q22_2 | The ideas I contributed were innovative— reflected original and creative ways to think of the problem. | {Strongly disagree}; {Somewhat disagree}; {Neither agree nor disagree}; {Somewhat agree}; {Strongly agree} |
| Q22_3 | The ideas I contributed were insightful— reflected deep and perceptive ways to think of the problem. | {Strongly disagree}; {Somewhat disagree}; {Neither agree nor disagree}; {Somewhat agree}; {Strongly agree} |
| Group | Source | Idea |
| M1 | AI | (High) Renewable Energy Projects: Investing in solar panels or wind turbines to generate clean energy for campus buildings. |
| Human | (Medium) Expand on the bike share program (more accessible) | |
| AI | (Low) Community Gardens: Creating spaces for students to grow their own food, fostering a sense of community and promoting sustainable agriculture | |
| M2 | AI | (High) Establish a student-led sustainability committee specifically focused on the dining hall, allowing students to actively participate in decision-making processes and initiatives related to sustainability in the dining hall. |
| AI | (Medium) Initiate a ‘Green Dorm Challenge’ where dorms compete to reduce energy and water consumption, with progress tracked and shared through a digital dashboard. | |
| Human | (Low) Card Swipe incentives: Have a particular amount of Flex dollars that can be earned for swiping at least a certain number of times in a semester to make sure dining hall workers can track and be prepared for students at meal times |
Human-AI Collaboration
Human-AI collaboration studies how people and AI systems divide, coordinate, and adapt work to support decisions, creativity, and learning.
See all 476 papers