Design can be understood as a distinctive mode of knowledge production in which knowing and making are mutually constitutive. Its epistemic specificity lies in its projective capacity to frame problems, construct representations, generate alternatives, and explore possible futures. Across architecture, urban design, landscape architecture, exhibition design, product design, and related fields, design therefore engages not only with what exists, but with what could be.

Generative artificial intelligence intensifies and reconfigures this condition. As computational systems increasingly participate in ideation, representation, simulation, variation, and evaluation, design knowledge emerges through distributed configurations of human and non-human agency.

The notion of Generative Design Ecologies describes these configurations as relational systems in which designers, models, datasets, archives, precedents, interfaces, prompts, regulations, and situated forms of expertise jointly condition the production of design possibilities.

From this perspective, generative design is not simply automated solution production, but the construction and exploration of spaces of possibility. Alternatives and scenarios become provisional knowledge objects that are generated, compared, tested, rejected, and recombined through iterative inquiry. The crucial issue is therefore the relationship between generation and judgement. While AI operates predominantly through probabilistic inference, design concerns the construction and evaluation of possibilities that are not yet actualised.

This tension foregrounds the continued importance of contextual interpretation, situated knowledge, ethical positioning, and responsibility. Generative processes are also inseparable from the epistemic infrastructures that enable them. Archives, datasets, classifications, precedents, and representational systems are not neutral repositories: they actively structure what can be retrieved, correlated, and generated. AI-augmented design thus involves a double operation: designing the knowledge ecology from which possibilities emerge, and designing through the possibilities that this ecology makes available.

This perspective directly addresses IFKAD 2027’s theme, The Ecology of Knowledge in the AI Era. Design disciplines provide a particularly relevant field for examining hybrid knowledge ecosystems because they combine explicit and tacit knowledge, analytical and aesthetic reasoning, technical constraints, and cultural values. At the same time, reliance on shared models and datasets may produce epistemic and aesthetic convergence, normalising statistically dominant patterns and creating new forms of design monoculture.

These transformations also concern organisational knowledge. Within design practices, AEC organisations, public agencies, creative firms, and universities, AI is changing how knowledge is curated, circulated, attributed, and reused, as well as how authorship, responsibility, and intellectual property are negotiated. Understanding these emerging generative design ecologies is essential if AI is to expand design’s exploratory capacity without displacing the diversity, judgement, and situated intelligence through which design produces knowledge, meaning, and value.