|
2nd General Meeting — Universidade de Vigo, Campus Crea, Pontevedra, 26–27 February 2026. 1. OverviewWorking Group 2 (Collective Intelligences) held its workshop session during the 2nd General Meeting in Pontevedra on 27 February 2026. The session was structured as a direct response to Michael Schwab’s keynote lecture, Local, Polylocal, delivered on Day 1, in which he proposed polylocality as a conceptual framework for understanding artistic intelligence. The WG2 workshop took up this framework as its central provocation, exploring how the concept of polylocality — the co-existence of multiple, non-reducible knowledge assemblages (“locals”) — might advance WG2’s mandate to investigate collective intelligences and co-agencies between human and non-human actors. 2. Alignment with WG2 Tasks and DeliverablesThe MoU charges WG2 with investigating how various efforts in art and culture that make use of machine learning and generative algorithms can contribute to shared responses to societal challenges. The Pontevedra workshop advanced all four of WG2’s defined tasks: |
|
WG2 Task |
MoU Description |
Pontevedra Progress |
|
T2.1 |
Prepare the ground for a common understanding of artistic intelligence as a response to AI |
Schwab’s polylocality framework was collectively tested as a foundation for understanding artistic intelligence. Participants positioned diverse practices — from interspecies performance to climate research to algorithmic cultures — in relation to this framework, building shared conceptual ground. |
|
T2.2 |
Identify the particular contribution of artistic and practice-based research |
The breakout group on “Verbing Polylocality” generated 22 verb-based “action activators” (e.g., listening, commoning, improvising, tracing) that articulate what artistic practice uniquely contributes: situated, embodied, process-oriented modes of knowledge production. |
|
T2.3 |
Consultations with scientific researchers, policy-makers, and stakeholders on the concept of artistic intelligence |
The workshop brought together researchers from art, design, music, computer science, philosophy, and cultural policy. Contributions addressed the intersection of AI with questions of spatial justice, copyright, open-source sustainability, and democratic governance. |
|
T2.4 |
Analyse the specific technological, institutional, and outreach capacities of stakeholders |
The introductions round mapped participants’ situated practices and institutional contexts. Three interlocking domains were identified as organising categories: technical architecture/infrastructure, policy/governance, and arts practice/aesthetics. |
|
The workshop’s outputs contribute directly to Deliverable 6: Artistic Intelligence Reference Framework (due M47), for which WG2 shares responsibility with WG3. The verb-based vocabulary developed in Pontevedra represents an early contribution to the Reference Framework’s stated aim of providing “a vocabulary supporting inquiries of artistic qualities, societal impact, and multiple value generation.” 3. Session Structure and Activities3.1 Context: Michael Schwab’s Keynote (Day 1)On the morning of Day 1, Michael Schwab (founding Editor-in-Chief of the Journal for Artistic Research) delivered the keynote Local, Polylocal. He proposed polylocality as an ontological framework for artistic intelligence, arguing that knowledge assemblages (“locals”) do not share a common ground outside of what is established through consultation and articulation. He hypothesised that context-density gradients between locals might help explain knowledge creation, preservation, and extractive knowledge systems — and that artistic research requires high context-density operations to meaningfully exist. This lecture set the conceptual agenda for WG2’s Day 2 workshop. 3.2 Introductions: Responding to PolylocalityThe workshop opened with a brief recap of themes from the 1st General Meeting in Rome (audience science, the power to be affected), before inviting each participant to position their research in direct dialogue with Schwab’s polylocality framework. This produced a rich mapping of how the concept resonates across disciplines. Key contributions addressed embedding and the materiality of data structures (Lucas Evers); delocalized knowledge operations and the thickening of nodes (Philippe Mairesse); the lived, fragmented nature of context and the problem of extraction (Kenan Zekic); the cycle between digital data and physical experience (Emily Hurdemann); interspecies performance and the limits of data capture (Bogdan Florea); locally embedded research in fire-prone multispecies landscapes and the promise of the polylocal for spatial and climate justice (Jo Scott); embodiment, post-humanism, and digital humanism (Marika Grasso); technical systems of communication as polylocal infrastructure (Yanai Toister); hyper-polylocality and hyper-polyphony as feminist, practice-based frameworks (Helena Barranha); platforming plural voices, pro-existence, and latent reality (Ayo Ayodele); swarm intelligence, social network analysis, and quantum computing as polylocal analogy (Monika Sonta); embodied AI and the imperative to bring artistic processes to bear on technology (Alessandra de Pisa); and the idea of treating AI as “literature” requiring multiple, situated and ethical perspectives (Kevin Walker). 3.3 Breakout GroupsThe workshop then split into two parallel groups:
3.4 ReconveningWhen the groups reconvened, Michael Schwab reflected on the discussion and highlighted the role of connectors as mechanisms for how locals might relate without collapsing into a single framework. He posed a question that the group recognised as a productive tension to sustain: are we discovering something new, or articulating something that has always been present but lacked a framework for recognition? 4. Key Outcome: “Verbing Polylocality”One substantial methodological output of the Pontevedra workshop was the development of a verb-based approach to polylocality. Rather than defining polylocality through fixed categories or taxonomies, this group proposed that it be understood as a set of operations — actions and practices that transpose the local into the polylocal. The resulting vocabulary of verbs was conceived as “action activators”: catalysts for practice that encapsulate ideas while also suggesting concrete research actions. This approach directly serves the Reference Framework (D6) by proposing a vocabulary that is practice-derived, process-oriented, and resistant to the flattening abstractions that participants identified as a risk in dominant AI discourse. The verbs were clustered into five thematic groups: |
|
Cluster |
Verbs |
Relevance to the Action |
|
Reception and Attunement |
Listening, Attuning, Sensitizing, Practising Curiosity |
Grounds the Reference Framework in embodied, multisensory modes of engagement. Oliveros’s Deep Listening was given as one example, underlining the Action’s focus on responsiveness. |
|
Exchange and Movement |
Exchanging, Call and Response, Traversing, Crossing, Echoing, Connecting, Resonating |
Addresses how knowledge moves across locals without losing context-density. Links to the Action’s core value of cross-fertilisation and its emphasis on transferability. |
|
Iteration and Experimentation |
Improvising, Repeating, Experimenting, Interpreting |
Articulates the iterative, non-linear character of artistic research. Connects to the Action’s commitment to practice-based, mission-driven approaches and the dynamic analytical matrix. |
|
Politics of Knowledge |
Commoning, Resisting, Holding a Space, Inhabiting, Opening |
Engages directly with questions of equity, accessibility, and responsibility — three of the Action’s four pillars. Raises critical questions about copyright, extraction, and the economics of knowledge production. |
|
Language and Worldmaking |
Mythmaking, Neologising, Tracing, Decontextualising / Recontextualising |
Proposes that the Reference Framework must be self-aware about its own language. Connects to the Action’s aim to develop vocabulary and frameworks that challenge dominant AI discourse. |
|
Florian Schneider’s closing synthesis framed the verb-based approach as a practice of decontextualising and recontextualising — an open-ended process whose results are always preliminary. The key insight: polylocality is fundamentally about process. No participant, practice, or framework is the endpoint; each is a node through which knowledge moves, transforms, and continues. 5. Cross-Working Group ConnectionsSeveral lines of inquiry that emerged in the WG2 workshop connect directly to the work of the other working groups:
6. Contribution to the Four Pillars of the ActionThe workshop engaged all four pillars of the Action: |
|
Pillar |
WG2 Contribution at Pontevedra |
|
Responsiveness |
The verb-cluster around listening, attuning, and sensitizing proposes responsiveness as a foundational practice for polylocal knowledge exchange. The insistence on reciprocity — no passive position — embodies the Action’s commitment to responsive engagement with digital technologies. |
|
Accessibility |
The call for transdisciplinary translations and a vocabulary accessible to non-researchers directly serves the Action’s accessibility principle. The critique of dominant AI vocabulary as exclusionary frames accessibility as a linguistic and epistemological challenge, not merely a technical one. |
|
Responsibility |
The discussion of extraction, colonisation of non-human knowledges, and the ethics of “upscaling” placed responsibility at the centre of the polylocality framework. Alessandra de Pisa’s insistence that AI is “too important to leave in the hands of the engineers” encapsulates WG2’s position. |
|
Equity |
The verbs of commoning, resisting, and holding a space engage directly with questions of who benefits from knowledge production. The discussion of copyright, open-source sustainability, and the economic structures that determine how resources are allocated connects equity to concrete material conditions. |
7. Next Steps for WG2Building on the Pontevedra workshop, WG2 will pursue the following priorities: Develop the context-density framework: Work with Schwab’s hypothesis that context-density gradients between locals can be analysed as a tool for understanding how artistic intelligence operates, contributing to the dynamic analytical matrix. Cross-WG collaboration: Coordinate with WG3 on the shared D6 deliverable, and engage WG1 and WG4 on questions of scale, infrastructure, and policy that emerged in the workshop. Stakeholder consultation: Continue consultations (T2.3) by engaging practitioners and researchers who were not present in Pontevedra with the polylocality framework and verb-based methodology, broadening the base of contributions. Sustain productive tensions: Hold open the question posed by Schwab — are we discovering or articulating? — as a generative tension rather than one requiring resolution, allowing it to inform the iterative development of WG2’s conceptual tools. |
This work is based upon work from COST Action Artistic Intelligence — Responsiveness, Accessibility, Responsibility, Equity (ARTinRARE), CA23158, supported by COST (European Cooperation in Science and Technology).
COST (European Cooperation in Science and Technology) is a funding agency for research and innovation networks. Our Actions help connect research initiatives across Europe and enable scientists to grow their ideas by sharing them with their peers. This boosts their research, career and innovation.