PaperlayerView exact revision on arXiv

Source-linked AI summary

AMINA: The Inclusive and Accountable AI for Marginalized Immigrant Nonprofit Assistance

Maryam Mokhberi, Dipto Das, Syed Ishtiaque Ahmed

arXiv:2608.30084v1cs.CYcs.HC

TL;DR

Iranian immigrant nonprofits face exclusion from formal systems and digital infrastructures, while misinformation and political constraints further shape their work. The paper studies these conditions through interviews, co-design, and prototype evaluation, finding that AMINA can support inclusive nonprofit coordination while reinforcing human collaboration, though its prototype does not establish AI-component efficacy.

  • Problem

    Iranian immigrant nonprofits face legitimacy barriers, capacity gaps, misinformation, and exclusion from formal registries and digital platforms.

  • Method

    The paper reports a three-phase mixed-methods study comprising 27 interviews, a co-design session, and seven evaluation and feedback interviews on an AMINA prototype.

  • Results

    AMINA was designed and evaluated as an inclusive AI assistant supporting routine operations, informal efforts, credibility, transparency, and collaboration.

  • Takeaways & Limitations

    The work positions AI as a collaborative partner that can strengthen, rather than replace, the human connections sustaining immigrant nonprofit practice.

  • Takeaways & Limitations

    The high-fidelity Figma prototype lacked backend integration, so findings concern perceived design value and legibility rather than AI-component efficacy.

Abstract

from arXiv · show

Immigrant-led nonprofit groups, particularly those operating in politically sensitive contexts, face exclusion from formal registries and digital platforms. This paper reports a three-phase mixed-methods study with Iranian immigrant nonprofit practitioners: 27 semi-structured interviews, a co-design session, and 7 evaluation and feedback interviews on a prototyped AI assistant, AMINA. Our findings highlight how legitimacy barriers, capacity gaps, and politically charged misinformation constrain nonprofit operations. We translate these insights into design goals for an inclusive nonprofit AI assistant: support for everyday group operations, recognition of informal nonprofit efforts, proactive countering of misinformation, and multilingual, accessible interaction. User evaluations show AMINAs potential to reduce reporting burdens and foster transparency through proactive reminders, and catalyze collaboration across dispersed networks. We contribute to CSCW and HCI by characterizing the cooperative work of transnational immigrant nonprofits, extending scholarship on informality and misinformation, and demonstrating how AI can act as a collaborative partner that strengthens, rather than displaces, the human connections at the core of nonprofit ecosystems, while also posing major risks.

1 Introduction

The paper examines how Iranian immigrant nonprofits operate across informal, politically sensitive contexts and proposes AMINA, an inclusive AI assistant designed to support their cooperative work without displacing human relationships.

  • Motivation: Iranian immigrant nonprofits face uneven AI access because complex infrastructures, missing formal status, and political barriers limit adoption and create potential harms.The study situates these challenges within broader inequalities affecting under-resourced and marginalized nonprofit groups.
  • Research gap: Prior HCI and CSCW research under-examines immigrant nonprofits whose cross-border cooperative work is shaped by distance, language, legal regimes, and geopolitical constraints.The paper frames these groups as sites of distributed cooperative work sustained through continuous articulation work.
  • Operational context: Messaging platforms support grassroots coordination by storing records, allocating volunteer tasks, and maintaining awareness, yet formal nonprofit systems structurally overlook these efforts.Informal groups use accessible digital tools because designated nonprofit platforms typically require formal registration.
  • Study approach: The three-phase study combines 27 practitioner interviews, a co-design session, and prototype evaluation interviews to investigate coordination and AI support for Iranian immigrant nonprofits.Interview findings identified technological exclusion, absent formal status, misinformation, and geopolitical bias as central design challenges.
  • Design direction: AMINA is designed to support routine nonprofit tasks, recognize informal efforts, address misinformation proactively, and foster inclusion, trust, collaboration, and sustainability.The assistant is positioned as a collaborative partner that augments rather than substitutes for transnational cooperative work.

2 Related Work

Related work shows that immigrant-led nonprofits sustain transnational charitable action through informal, resource-constrained cooperation, while existing nonprofit technologies and AI research inadequately address their political and infrastructural conditions.

  • Immigrant identity and transnational practice: Immigrant charitable work supports cultural identity, belonging, and transnational ties, but informal groups often lack recognition, funding, and access to nonprofit technologies.These exclusions are linked to immigration status, cultural and linguistic marginalization, and transnational political tensions.
  • Organizational constraints: Immigrant-led nonprofits differ from stable bureaucratic organizations because volunteer labor, informal governance, limited resources, and cross-border pressures shape their everyday operations.They rely heavily on social capital and continuous maintenance of organizational routines, communication systems, and networks.
  • Cooperative work: CSCW concepts such as articulation work, mutual awareness, and common information spaces explain how dispersed participants coordinate interdependent charitable tasks.Groups build protocols and shared artifacts to reduce the work required to align actors, resources, and activities.
  • Informal organizing: These groups use adaptive organizing practices to balance visibility, credibility, autonomy, and safety when formal institutional pathways are inaccessible, risky, or misaligned.Alternative organizing emphasizes collective labor, flexible governance, and improvised structures rather than disorganization.
  • Technology and AI: AI-as-collaborator research largely studies well-resourced professional teams, leaving volunteer-run, resource-poor, politically vulnerable collectives underexamined.Existing nonprofit tools may reflect design fictions, add administrative burdens, or drive mission drift through data demands.
  • Research gap: The paper addresses this gap by focusing on inclusive systems attentive to migration-related precarity, geopolitical constraints, and informal infrastructures.It centers group-organized practices of immigrants operating in politically sensitive contexts.

3 Methods Overview

The study uses a three-part mixed-methods design spanning interviews, co-design, and prototype evaluation with Iranian immigrant nonprofit participants across several regions.

  • Study design: The study comprises semi-structured interviews, a co-design session, and evaluation and feedback sessions on the AMINA prototype.The phases examine nonprofit technology use, inform design, and assess the prototype.
  • Recruitment: Recruitment used flyers on the first author’s Instagram and Telegram channels followed by snowball sampling to reach informal and transnational charitable actors.This approach targeted participants difficult to identify through institutional recruitment channels.
  • Analysis: All sessions were conducted in Farsi and analyzed through inductive thematic analysis involving independent coding, consensus meetings, a maintained codebook, and affinity-diagram clustering.Themes were consolidated across phases, with Phase 1 findings serving as sensitizing concepts for later phases.

4 Phase 1: Understanding Charitable Activities in the Iranian Immigrant Population

Iranian immigrant nonprofits sustain distributed charitable work despite legitimacy barriers, limited capacity, and politically charged misinformation. Participants valued AI’s operational potential but cautioned that systems built on exclusionary infrastructures could reproduce existing harms.

  • Technology opportunities and risks: Participants saw AI and information technology as valuable for reducing operational labor, but warned against deploying data-driven systems on exclusionary and misinformed digital infrastructures.Reported uses included drafting documents, accounting, receipts, donor profiles, and reporting, while volunteer-built systems could be difficult to maintain.
  • Cooperative charitable work: IING charitable campaigns coordinate diaspora leaders, volunteers, donors, Iranian partners, and peer organizations across distance, time zones, and voluntary divisions of labor.These arrangements support accountability, beneficiary identification, aid delivery, donor communication, and joint projects.
  • Legitimacy barriers and structural exclusion: Formal registration barriers tied to international politics, sanctions, scarce resources, and volunteer capacity exclude many groups from nonprofit directories, fundraising platforms, banking services, and donation matching.Groups rely on cash or personal accounts and intermediaries, but these workarounds constrain scale and sustainability.
  • Capacity gaps: Capacity gaps arise from limited resources, language and cultural barriers, digital literacy differences, trust and safety concerns, and high volunteer turnover.Volunteer departures can take organizational knowledge with them, disrupting continuity and technology adoption.
  • Misinformation and political stigmatization: Misinformation about political affiliations or aid recipients creates reputational harm, internal distress, and pressure to reduce public visibility or intensify transparency practices.Groups reported false accusations on social media and messaging platforms, while detailed reports and documentation consumed labor and sometimes limited growth.
  • Digital invisibility: Informal and under-resourced groups often lack online traces, making impactful work digitally invisible and less discoverable by emerging AI systems.This invisibility follows from tools and services that assume formal registration while overlooking capacity gaps.

5 Design Goals

The design goals translate interview findings into an AI assistant that supports nonprofit operations while recognizing informal efforts and addressing misinformation. The proposed system is intended to serve both nonprofit workers and donors or audiences.

  • Design approach: The design approach is guided by design justice, positioning immigrant nonprofit workers and their communities as those who should lead the design.The system also supports everyday care practices extending beyond formal organizations.
  • Intended users and scope: The proposed system serves donors and audiences through cause discovery, donations, contribution tracking, and impact reports, while supporting managers and volunteers with operational and organizational tasks.The design is framed as inclusive of informal practices and groups with capacity gaps.
  • DG1: Support operation needs: The assistant should centralize fundraising, donation, receipt, impact-reporting, campaign, event, and coordination tasks in an accessible platform.This goal responds to workers’ reports that operational tasks consumed time needed for community service.
  • DG2: Recognize informal charitable efforts: The system should validate and support charitable efforts regardless of legal registration status, enabling informal groups to participate in broader nonprofit ecosystems.This goal addresses the financial, managerial, and information-technology disadvantages associated with lacking official nonprofit status.
  • DG3: Be accountable for misinformation: The assistant should acknowledge and act toward online misinformation, particularly misinformation shaped by political biases against Iran or groups connected to Iranian communities.The goal follows participants’ reports that misinformation posed severe challenges to nonprofit work.

6 Phase 2: Co-Design Session

Phase 2 used participatory co-design to translate interview findings into concrete expectations and features for an AI assistant supporting Iranian immigrant nonprofits. Participants emphasized operational coordination, transparency, accessibility, privacy, trust, and recognition of informal groups.

  • Session approach: Participants used expectation mapping, persona construction, and design ideation to translate high-level design goals into actionable AI-assistant features.The nine-participant co-design session included nonprofit executives, volunteers, and frequent donors.
  • Needs and constraints: Donors sought transparent missions, beneficiary criteria, financial reporting, and coherent communication, while executives described reporting and coordination as labor-intensive.These expectations created a persistent mismatch between donor demands and available time, labor, and coordination capacity.
  • Inclusive access: Participants identified accessibility and multilingual support as essential for users with limited digital literacy, disabilities, or diverse language backgrounds.The envisioned users included an elderly Afghanistani woman, a second-generation Iranian immigrant, and a blind Turkish-speaking student.
  • Trust and privacy: Privacy and trust were central concerns for donors with unstable immigrant status or ties to politically complex regions.Participants wanted assurance that conversations, identities, data, and transactions would remain secure and undisclosed to governments.
  • Legitimacy and misinformation: Participants proposed trust-network endorsements and community verification as alternatives for recognizing informal nonprofit legitimacy.These mechanisms included endorsements from friends or established nonprofits and multi-criteria rating systems.
  • Legitimacy and misinformation: They also proposed expandable fact-checking elements, voice interaction, translation, and simplified buttons to counter misinformation and reduce capacity barriers.Fact-checking elements would provide documentation such as financial reports, images, or videos for verifying nonprofit activities and affiliations.

7 Phase 3: Designing AMINA: AI for Marginalized Immigrant Nonprofit Assistance

Phase 3 translated the co-design findings into AMINA, a conversational AI assistant with public and nonprofit-facing components. Its workflows combine nonprofit discovery, credibility and reporting support, coordination, planning, and accessible interaction across text and voice.

  • Prototype development: AMINA was prototyped from the first two study phases to support nonprofit operations while incorporating design elements addressing inclusion, legitimacy, misinformation, and capacity gaps.The prototype was built on Figma and refined through iterative sessions.
  • System components: The Explorer Agent answers public questions using formal directories, web and social-media data, community metadata, and optional nonprofit inputs.It also provides profile-management dashboards, particularly for informal groups with limited digital presence.
  • System components: Local Agents integrate private or semi-private nonprofit data to answer questions about donations, campaigns, volunteering, and organizational documents.These nonprofit-specific instances can be deployed on organizational websites or social-media channels.
  • Donor and public workflows: Donor workflows support discovering and comparing registered or informal nonprofits, exploring profiles, donating, volunteering, reviewing groups, and tracking interactions.Profiles include mission, campaign, financial, misinformation-response, and credibility information.
  • Nonprofit workflows: Organizer workflows support profile management, AI-assisted event planning, structured report generation, and proactive disclosure intended to pre-bunk misinformation.Reports can be quantitative or qualitative, while disclosures may cover affiliations, beneficiaries, funding origins, and expenditures.
  • Coordination: AMINA acts as a coordination mechanism through shared dashboards, planning tools, meeting notes, and reminders about task states.These features aim to preserve organizational continuity across volunteer turnover and reduce coordination work distributed across individuals.

8 Users’ Feedback on AMINA

Users valued AMINA’s support for informal nonprofit legitimacy, operational work, misinformation response, accessibility, trust, transparency, and collaboration, while identifying risks around bias, misuse, privacy, and interface design.

  • Recognizing informal nonprofits: Community verification tags and multi-criteria ratings offered alternatives to government registration for recognizing informal nonprofits.Participants viewed these mechanisms as ways to acknowledge community-based groups, but questioned whether visual tags could introduce bias and whether numeric scores suit humanitarian work.
  • Recognizing informal nonprofits: Participants warned that alternative legitimacy mechanisms could enable scams or money laundering without safeguards.They suggested that AI might help detect suspicious behavior, highlighting a tension between inclusion and new risks.
  • Addressing misinformation: Five participants supported prominent proactive misinformation debunking, while two cautioned that it could unintentionally undermine trust.Suggested alternatives included a subtler panel or conversational clarification on request.
  • Accessibility and human oversight: Participants requested improved Farsi interaction, customizable and less text-dense interfaces, and voice-only options for varied accessibility needs.They also emphasized human escalation and volunteer-in-the-loop mechanisms when the AI cannot fulfill requests.
  • Proactivity and trust: Proactive reminders for receipts, photos, meeting minutes, and coordination summaries were valued for strengthening transparency and organizational sustainability.Participants also wanted clear privacy assurances for sensitive beneficiary data and legal questions.
  • Collaboration and capacity building: Participants envisioned AMINA as a network hub that connects dispersed nonprofits, supports knowledge-sharing, and helps volunteers overcome cultural, linguistic, and digital barriers.They proposed collaboration across organizations and proactive outreach to emerging or disconnected initiatives.

9 Discussions

The discussion frames informal immigrant nonprofit practices as community expertise shaped by exclusion, while positioning AI as a cooperative partner that must balance inclusion, misinformation response, and accountability.

  • Proactivity and Nudging: Proactive reminders and feedback can sustain volunteer-run groups in low-resource, politically sensitive nonprofit contexts.The study defines nudging as subtle interventions that guide behavior through reminders and feedback.
  • Informal Practices in HCI: Informal infrastructures are community innovations shaped by geopolitical exclusion, sanctions, and platform erasure rather than merely gaps awaiting formalization.Design justice therefore starts with what already works at the community level and scaffolds it without overwriting community-defined practices.
  • Informal Practices in HCI: Informality enables continuity and trust for immigrant-led nonprofits when registries, banking systems, and official data sources are inaccessible.The paper characterizes these arrangements as survival strategies and politically situated resistance to systemic exclusion.
  • Accountability and Risk: Extending legitimacy to excluded informal groups can also extend legitimacy to scams and money laundering, creating a trade-off between gatekeeping and donor risk.Graduated trust and AI-assisted anomaly detection are proposed as ways to soften, but not eliminate, this tension.
  • Misinformation: Prebunking is particularly important because politics-related misinformation can directly undermine donor trust in immigrant nonprofit contexts.How prebunking appears in the interface matters, and its placement becomes part of the design problem.
  • AI as a Collaborative Partner in Nonprofit Work: AMINA supports cooperative work by absorbing articulation tasks such as accounting, report assembly, and organizational memory across volunteer turnover.Its value extends beyond individual task automation by helping preserve arrangements constrained by scarce volunteer labor.
  • AI as a Collaborative Partner in Nonprofit Work: AMINA mediates asymmetric donor and organizer needs through reports, ratings, and prebunking panels that function as contested boundary-negotiating artifacts.Phase 3 disagreement about prebunking prominence illustrates that transparency displays require continuing negotiation rather than one-size-fits-all standardization.
  • AI as a Collaborative Partner in Nonprofit Work: Volunteer-in-the-loop collaboration makes AI’s value partly depend on routing privacy-sensitive and politically fraught decisions back to humans.The assistant’s role is therefore qualified by knowing what not to handle.

10 Limitations

The study’s generalizability is constrained by its geographically and demographically narrow sample, excluded stakeholder perspectives, and network-based recruitment. The prototype supports design feedback but cannot establish AI-component efficacy.

  • Scope and Sampling: The sample primarily represents Iranian diaspora communities in North America and Europe, leaving several other regions underrepresented.East and South Asia, Turkey, Arab countries, and Australia were specifically underrepresented.
  • Scope and Sampling: The study focused on Iranian immigrants, donors, and volunteer organizers, excluding other diasporas and perspectives from beneficiaries, state actors, and intermediary nonprofits.These boundaries limit how broadly the findings can be generalized across transnational philanthropic contexts.
  • Scope and Sampling: Recruitment through the first author’s networks and participant referrals may have produced network-based sampling bias.Connected individuals may have been more likely to participate, shaping perspectives on political sensitivities, misinformation, and legitimacy.
  • Interpretive Boundary: The findings should be read as situated accounts of recurring challenges and design needs rather than statistically representative estimates of the broader Iranian diaspora.
  • Prototype Boundary: Because AMINA was a high-fidelity Figma mock-up without backend integration, the evaluation addresses perceived value and legibility rather than AI-component efficacy.Assessing efficacy requires a functional deployment.

11 Future of the Work and Design Implications

Future work should strengthen AMINA’s proactive support while preserving volunteer agency and incorporating empathy into subsequent iterations.

  • Design Implications: Future iterations should strengthen proactive reminders for transparency, organizational sustainability, privacy assurance, and misinformation prebunking.These reminders should augment rather than replace volunteer labor in cooperative nonprofit work.
  • Design Implications: Future versions should incorporate human-like qualities because empathy is important in nonprofit work.

12 Conclusion

The paper examines how Iranian immigrant nonprofits navigate legitimacy barriers, fractured infrastructures, and politically charged misinformation, then designs and evaluates AMINA as an inclusive AI assistant. It concludes that justice-oriented AI can support credibility and transparency without excluding informal efforts while strengthening human connections.

  • Conclusion: The study traces how legitimacy barriers, fractured infrastructures, and politically charged misinformation shape Iranian immigrant nonprofit work.
  • Conclusion: The authors designed and evaluated an AI assistant prototype addressing nonprofit and donor needs while supporting credibility and transparency without excluding informal efforts.
  • Conclusion: The paper positions immigrant-based nonprofits as critical sites of care work and advances justice-oriented AI design that strengthens rather than replaces human connections.

A Appendix: Visuals of AMINA Interface Components

The appendix presents AMINA’s prototype interface components, spanning nonprofit discovery, donor interaction, transparency, project management, community evaluation, volunteering, and multilingual access.

  • Discovery and recognition: AMINA’s landing page and Explorer Agent combine an AI chat interface with nonprofit listings labeled “Community Verified” or “Government Registered”.These elements present community-based and government-based recognition within the nonprofit discovery experience.
  • Donor interaction: The Local Agent page provides a chat interface with simplified action buttons tailored to donor needs.The prototype also includes a donor profile showing impacts, profile information, and transaction history.
  • Misinformation response: AMINA includes a debunking misinformation tab with fact-checking controls, claim and reality fields, and sections for supporting evidence.The design encourages nonprofits to proactively document and counter viral misinformation.
  • Organizational support: Additional components support community evaluation, volunteer hiring, executive project management, and Farsi translation.Together, these visuals cover evaluation, recruitment, organizational coordination, and multilingual interaction.
  • Transparency: Reports and transparency pages present qualitative stories and quantitative reports as separate interface tabs.These pages organize nonprofit reporting through narrative and quantitative views.
Loading 2608.30084v1…