AAAI-26 Community Activities

Sponsored by the Association for the Advancement of Artificial Intelligence
January 22-25, 2026 | Singapore EXPO | Singapore

Community Activities Schedule

Description of Activities

Applied Generative AI in Industry: Demos and Discussions from the Field

This 1–2 hour in-person community activity explores how Generative AI–based strategies—including large and small language models (LLMs and SLMs), fine-tuned domain models, retrieval-augmented generation (RAG) pipelines, and integrations with graph and vector databases—are applied across industries to address practical challenges.

Through live demonstrations and expert discussions, participants will examine how tools such as but not limited to LangChain, LlamaIndex, LangGraph, and modern data systems (e.g., Pinecone, Weaviate, FAISS, and Neo4j) enable reasoning- and action-based AI workflows that are scalable, modular, and context-aware. The session bridges research and real-world implementation, illustrating how agentic architectures coordinate multi-step reasoning, decision-making, and tool interaction across enterprise contexts.

The workshop emphasizes that no single approach is universally optimal—the right strategy depends on the framework, data characteristics, and the evaluation lens. Participants will explore how to design explainable and adaptive systems that align model capabilities with domain needs. For instance, when fine-tuning may outperform retrieval-based adaptation; when chunking and hybrid search strategies enhance recall and relevance; or how weighted semantic ranking and hybrid retrieval can maximize the utility of existing data—regardless of scale or structure.

The session will also cover data-enriching techniques (e.g., synthetic data generation, retrieval-based augmentation, and domain adaptation) alongside robust evaluation frameworks for assessing reasoning, reliability, and interpretability.

Key themes include framework and architecture selection, fine-tuning vs. retrieval trade-offs, data preprocessing strategies, evaluation methodologies, and responsible deployment. The goal is to empower participants with the insight to build explainable, high-performance generative and agentic AI systems that leverage any data effectively—from prototype to production

Topics

• Applied generative AI
• Large and small language models
• Retrieval-augmented generation (RAG)
• LangChain and LlamaIndex workflows
• Fine-tuning and domain adaptation
• Ethical and responsible AI
• Use cases in finance, manufacturing, and science.

Format of Community Activity: Duration: 1–2 hours.

Format

Interactive demos followed by a multi-sector expert panel.
Demo session (approx. 45–60 minutes): participants present short demonstrations using frameworks such as LangChain, LlamaIndex, OpenAI or open-source APIs, fine-tuned models, and RAG architectures to illustrate how different stacks perform in distinct contexts. Any new approach is highly appreciated.

Panel discussion (approx. 45–60 minutes): industry experts from finance, manufacturing, and science discuss adoption trends, infrastructure trade-offs, and lessons from real deployments. The session encourages open discussion, practical takeaways, and cross-sector learning.

Attendance

Expected 40–60 participants, including students, researchers, and professionals. Open attendance; no prior experience with generative AI required.

Submission Requirements

Participants are invited to submit short demo proposals (1–2 pages) describing the use case, AI framework or architecture used, and insights into what worked or did not; optional short videos or repository links may be included.

Submission Site Information

Submissions and inquiries: komalssharan@gmail.com or [easychair link]

Organizers: TBD; contact:  Komalssharan@gmail.com , Komal Sharan

Bias in Multimodal AI: Representation, Risk, and Repair

The Bias in Multimodal AI: Representation, Risk, and Repair workshop aims to advance understanding of how bias emerges, propagates, and can be mitigated in multimodal AI systems. We invite researchers, practitioners, and policy experts to submit work that explores bias from technical, social, and ethical perspectives, fostering dialogue toward fairer and more inclusive AI.

Topics

Submissions may include, but are not limited to:

• Bias propagation across modalities (text, image, audio, time-series)
• Case studies of stereotypical outputs and misrepresentation in MLLMs
• Auditing frameworks and benchmark datasets for multimodal bias evaluation
• Inclusive and representative dataset construction for foundation models
• Prompt engineering and parameter tuning for bias reduction
• Community-driven and participatory design for responsible multimodal AI
• Explainability and transparency methods for bias detection
• Ethical and policy considerations for fair deployment of multimodal AI
• Multidisciplinary and cross-cultural perspectives on representation and fairness

Format of Community Activity

• Invited Talk (1.5 hours): delivered by internationally recognized experts in LLMs, CV, or time-series learning to share insights on current multimodal learning capabilities and research frontiers within bias detection and management. (Prof. Eduard Hovy, University of Melbourne and Carnegie Mellon University)
• Contributed Paper Presentations (3 hours) where selected non-archival submissions will be presented to showcase ongoing multimodal bias detection and management work.
• Concluding Panel Discussion (1.5 hours) bringing together speakers and organisers to engage the community on the future of detecting and managing bias through multimodal intelligence.

Attendance: 50 to 80 participants.

Submission Requirements

We welcome non-archival submissions in the following formats:

• Short Papers (5 – 9 pages): Empirical findings, datasets, or technical analyses.
• Long Papers (10 pages): Empirical findings, datasets, or technical analyses.
• Position Papers (up to 3 pages): Conceptual arguments or perspectives on representation and equity.
• Application Papers (up to 4 pages): Applied or industry examples of bias detection and repair.

All submissions must be in PDF format and use the CEUR-WS template. Accepted papers will be presented at the workshop and included on the official AAAI Community Activities website. The selected papers will be published at CEUR-WS proceedings.

Community Activities Committee

• Soyeon Caren Han, University of Melbourne, Australia
• Rina Cabral, University of Melbourne, Australia
• Eduard Hovy, University of Melbourne, Australia
• Josiah Poon, University of Sydney, Australia
• Luca Cagliero, Politecnico di Torino, Italy
• Honghan Wu, University of Glasgow, U.K.
• Sangwook Yi, Hanyang University, South Korea
• Kyungreem Han, KIST, South Korea
• Seungryong Kim, KAIST, South Korea
• Kate Shim, KAIST, South Korea
• Minseok Song, POSTECH, South Korea 
• Yihao Ding, University of Western Australia, Australia
• Goran Nenadic, University of Manchester, U.K.
• Christopher Leckie, University of Melbourne, Australia
• Prasenjit Mitra, Carnegie Mellon University, Rwanda

Community Activities External URL: https://sites.google.com/view/aaai26-biasin

Empowering Global South AI Talent: Bridging Gaps for Underrepresented Researchers in AI

This AAAI-26 Community Activity aims to empower researchers and students from the Global South by addressing disparities in AI education, mentorship, and global collaboration. The two-hour, in-person session will bring together students, early-career researchers, and academics for mentoring, networking, and skill development in AI research. The activity will feature a keynote talk, a panel discussion on inclusive research ecosystems, and a 3-Minute Thesis (3MT) competition for selected abstracts.

Participants will engage with leading experts and gain exposure to topics in efficient deep learning, computer vision, NLP, and AI for social good. The session also provides a platform to discuss actionable strategies for increasing participation and equity in global AI research communities. Researchers from underrepresented backgrounds, particularly women, early-career scientists, and participants from low-resource institutions, are strongly encouraged to participate.

Topics

Practical AI for resource-constrained environments; efficient deep learning and model compression; deployment on low-power and edge devices; coreset selection for fair and data-efficient training; generative methods for anomaly detection; computer vision for low-resource settings, including image recognition, scene understanding, and visual quality assessment; NLP for multilingual and low-resource languages, equitable language technologies, and AI-driven governance tools; socioeconomic and gender disparities in AI; cultural factors in technology adoption; data-efficient learning paradigms; remote sensing for sustainable development; algorithmic fairness and bias mitigation; applications in healthcare, agriculture, education, environmental monitoring, disaster management, climate, and maritime sectors

Format of Community Activity

Two-hour in-person interactive session including a keynote address, a panel discussion, and a 3-Minute Presentation (3MT) competition for selected abstracts. The event encourages open discussion, mentoring, and networking among global participants.

Attendance

Open to all AAAI-26 attendees. Priority engagement for students, early-career researchers, and participants from underrepresented or low-resource regions. Expected attendance: 50-100 participants.

Submission Requirements

1–2 page anonymized PDF abstract (non-archival) following the AAAI-26 format. Submissions may include technical contributions, community initiatives, or case studies. Selected abstracts will be presented in the interactive session or 3MT competition.

Submission Site Information

Submissions via OpenReview: Empowering Global South AI Community Activity Workshop (AAAI-26)  
Website: https://sites.google.com/view/globalsouthai-aaai-26

Community Activities Committee

• Tushar Shinde (Chair) – Assistant Professor, IIT Madras Zanzibar, Tanzania, shinde@iitmz.ac.in  
• Patrick Le Callet – Professor, Université de Nantes / Polytech Nantes, France
• Maria G. Martini – Professor, Kingston University London, UK
• Mohammed El Hassouni – Professor, Mohammed V University, Morocco
• Monojit Choudhury – Professor, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), UAE
• Nakul Jain – CEO, Wadhwani AI Global, India
• Bibhas Adhikari – Fujitsu Research of America, Inc., USA
• Sunayana Sitaram – Principal Researcher, Microsoft Research India
• Kumar Rahul – Senior Research Scientist, Amazon Prime Video, USA
• Jonathan Shock – Associate Professor, University of Cape Town, South Africa
• Lim Chern Hong – Senior Lecturer, Monash University Malaysia
• Divya Sharma – Assistant Professor, York University, Canada
• Ankit Agarwal – Assistant Professor, University of North Carolina at Charlotte, USA
• Avinash Kumar Sharma – Researcher, IIT Madras, India
• Aisha Hamad Hassan – Researcher, IIT Madras Zanzibar, Tanzania

Contact Email: globalsouthai-aaai-26@googlegroups.com

Important Dates

• November 5, 2025 – Abstract Submission Opens
• November 20, 2025 – Abstract Submission Deadline
• December 1, 2025 – Notification of Acceptance (Rolling)
• January 20–27, 2026 – [TBD] at AAAI 2026, Singapore

Travel Funding

Travel support (subject to sponsor funding) is available for students and early-career researchers, with priority given to participants from underrepresented groups and low-resource institutions. Awards may also cover registration fees.
How to Apply: During OpenReview submission, select “I would like to be considered for financial aid” and provide a brief justification.

Participants may also apply for the AAAI Student Scholar Volunteer Program, which offers additional travel assistance for full-time undergraduate and graduate students. More information and application: AAAI Student Scholar Volunteer Program

Call For Sponsors

We welcome organizations and individuals to support travel grants, 3MT Competition Prizes and workshop activities, helping empower underrepresented AI researchers. Sponsors will be recognized on the website, during the event, and in all workshop materials.

Ethics of Example: Raising AI through Everyday Interactions

Artificial Intelligence is often seen as something to design and regulate. Ethics by Design remains essential, yet design and governance evolve linearly while technology advances exponentially, creating a persistent ethical gap.

This interactive workshop introduces a complementary framework: Ethics of Example. It explores how AI is not only engineered but also “raised” through billions of daily human-AI interactions. Every question, every tone, every act of patience or impatience teaches AI something about human nature, shaping its trajectory in real time.

We call this the mirror effect: AI reflects back the qualities we exhibit. If we approach with frustration, it amplifies frustration. If we engage with respect, it amplifies respect. Human agency is immediate and continuous.

This perspective aligns with values of education, shared responsibility, and collective growth, resonating with Singapore’s kampong spirit, where individual actions contribute to collective well-being. Just as communities thrive when each member is mindful of their impact, our digital ecosystem benefits when we approach AI interactions with awareness and intention.

Drawing on Geoffrey Hinton’s recent call for “AI with motherly instincts,” we examine who is already teaching AI these instincts: mothers, educators, healthcare professionals, and communities practicing ethics of care are the invisible AI educators shaping tomorrow’s intelligence.

Topics

AI ethics; responsible AI; human-AI interaction; ethics of care; behavioral AI education; AI literacy; conscious technology engagement; human agency in AI development; community-driven AI ethics.

Format

2-hour interactive workshop:

• 15-minute introduction to Ethics of Example and the mirror effect
• Three hands-on exercises (30 minutes each) using the EducatingAI Toolkit: 4-Second Check (mindful pausing), Mirror Test (reflecting on tone and intent), Reframe Exercise (transforming commands into collaborative requests)
• 15-minute collective reflection and comparison

Participants work in small groups of 6-8, with each group receiving different prompt variations for the same AI tasks, enabling real-time demonstration of how interaction styles affect AI responses.

Attendance

35-40 participants expected. Open to all AAAI-26 registered attendees curious about their role in shaping AI. Everyone welcome: students, researchers, industry professionals, educators, and community members. No technical background required: just curiosity and willingness to reflect.

Submission Requirements

None. Open participation for all registered AAAI-26 attendees.

Contact: nicoletta@womeninai.co

Organizer: Dr. Nicoletta Iacobacci, Founder, EducatingAI, Switzerland

External URL: nicolettaiacobacci.com

From Lab to Learning: Translating AI and Robotics Research into K-12 Outreach 

This workshop introduces a six-stage framework for translating AI research into engaging K-12 educational experiences that reach demographically diverse audiences. The framework guides participants through identifying translatable research concepts, selecting age-appropriate platforms, designing scaffolded curriculum, training facilitators, partnering with community organizations, and implementing data-driven improvements. Using the Bot Blitz case study, participants will observe Sphero RVR+ robots navigating narrative-driven challenges adapted for different grade levels while learning to make advanced AI concepts accessible to young learners.

https://athena.duke.edu/education-outreach/eaai-workshop

Format

Interactive workshop with three components: (1) 20-minute presentation detailing the six-stage framework; (2) 10-minute live demonstration using Sphero RVR+ robots navigating Bot Blitz challenges while explaining framework application; (3) 45-minute hands-on application workshop where participants work in small groups to apply the framework to their own research, develop preliminary outreach plans using provided templates, and identify strategies for building equitable community partnerships. Length: 75 minutes.

Attendance

Expected 15-20 participants. Open to all conference attendees including academics, graduate students, outreach coordinators, and educators interested in STEM education. No application or qualification criteria required.

Community Activities Committee

• Sandra Roach, Duke University, sandra.roach@duke.edu
• Karis,Boyd-Sinkler, Duke University, karis.boydsinkler@duke.edu
• David Hunt, Duke University, david.hunt@duke.edu
• Visrut Sudhakar, Duke University, visrut.sudhakar@duke.edu
• Shaundra Daily, Duke University, shani.b@duke.edu
• Miroslav Pajic, Duke University, miroslav.pajic@duke.edu
• Whitney McCoy, Duke University, whitney.mccoy@duke.edu

How to Do AI Research That Matters

This AAAI-2026 community activity will provide inspiration, practical advice, and guidance to PhD students (and early–career researchers such as postdocs) working in AI/ML research.

Objectives

• Broaden perspectives on ambitious and impactful AI research, including historical, societal, methodological, and ethical dimensions.
• Reinforce good practices: problem framing; rigorous experimental design; reproducibility and integrity; learning from failure; resilience.
• Expose students to perspectives from academia, industry, and government about what “impact” means and how it is pursued.
• Enable two-way interaction: allow students to ask questions, share concerns, and shape part of the content via a pre-event survey and open discussion

Topics

This event will focus on broader discussions about AI research:

• How to choose research problems and shape research ideas
• How to produce good science
• How to navigate the field, in academia and industry
• How to build a meaningful research trajectory

Format of Community Activity

The opening and core sessions will be delivered jointly by the organizers of this community event.

• Opening (45 min): “The History of AI Research & What It Means to Do Ambitious, Impactful Research.”
• Core Sessions (2 × 30 min):
  • “From Ideas to Experiments”: Effectively framing problems, rigorous evaluation, and learning from failures.
  • “From Experiments to Papers”: Writing strong papers in the era of foundation models, reproducibility, and reporting of empirical results
• Panel (45 min): Invited panelists from academia, industry, and government discuss their views on impactful research and advice to PhD students.
• Open Discussion (45 min): Audience–driven Q&A and dialogue, informed by a pre-conference survey.
• Refreshments / Networking: Coffee, tea, snacks, and light food to foster informal interactions.

Attendance

This event is open to all AAAI attendees, and is primarily targeted at early-stage researchers (PhD student and postdocs) as well as pre-PhD students interested in AI research. We expect ~100 participants.

Submission Requirements

This event does not require preliminary submission, but attendees are encouraged to answer the following survey: https://forms.gle/99CES1Faz7SxYriXA

Event Website:

https://sites.google.com/view/ai-research-that-matters?usp=sharing

Organizers

Stefano V. Albrecht, NTU Singapore
Reuth Mirsky, Tufts University
Manuela Veloso, Carnegie Mellon University
Matthew E. Taylor, University of Alberta

Teacher and Students-Building in AI Workshop

This interactive, half-day workshop is designed to forge a collaborative bridge between AI education theory and hands-on practice. Our primary objectives are to empower K-12 educators with immediately usable strategies for integrating AI into their classrooms and to engage students directly with creative, age-appropriate AI-building activities. We champion a “building together” philosophy, fostering essential dialogue and shared learning between educators and learners.

Key topics include demystifying AI literacy fundamentals, exploring pedagogical integration across subjects, and introducing hands-on tools like DeepSeek, ChatGPT and Gemini. A core focus is on embedding ethical design and critical thinking into student projects from the start.

The 4-hour format is highly participatory, blending brief presentations with extended collaborative work. It features an opening panel with diverse voices, followed by parallel tracks: one for teachers to deepen pedagogical skills and another for students to dive into guided AI use for study. The core of the workshop is a collaborative build session where mixed small groups of teachers and students work together on a shared challenge, culminating in a gallery walk to showcase projects and share insights.

Attendance is limited to 80 participants to ensure a quality experience: 30 educators and 50 middle/high school students (grades 7-12). Educator attendees are selected based on a brief statement of interest. Those wishing to facilitate a session may submit a 1-page activity abstract.

Submission & Committee

Educator statements and facilitator abstracts can be emailed to eden.mucache@gmail.com

The committee includes Dr. Eden Sansao Mucache (AI Consultant, ISCTEM School) and Dra. Milene Fausta Vilhete(Maputo Central Hospital, AI consultant)

For the detailed agenda, resources, and facilitator bios, please visit: https://www.amosa.org.mz/