Interactive genetic algorithms for architecture: evaluating the capacities of interaction modes to integrate designers’ intuitive judgments

Purpose This research investigates how human interaction with the interactive genetic algorithm (IGA) influences the formation of design outcomes. It aims to: (1) assess designers’ perceptions of the effectiveness of three different interaction modes of IGAs in achieving subjective and diverse outcomes, (2) understand how interaction patterns influence the pursuit of subjective goals and (3) examine differences between novice and expert designers. Design/methodology/approach A design experiment with 26 international participants used Snowflake, a multi-mode IGA tool, to evolve urban morphologies. Post-experiment surveys and interviews were conducted to address six research questions. Quantitative and qualitative analyses were employed to examine participants’ preferences, perceptions and interaction patterns. Findings Results showed that interaction modes influenced the evolutionary process in distinct ways and revealed differing perspectives, biases and factors that impacted evaluations across varying levels of design experience. Research limitations/implications Due to the small sample size and focus on the conceptual design phase, generalizability remained limited. Future studies are encouraged to extend these propositions to later design stages and broader contexts. Practical implications IGAs can reshape professional practice and design pedagogy by enabling architects and students to engage with design as a process of structuring problems and objectives rather than generating final forms. Through defining and interacting with the system, designers retained agency over the evolutionary direction of outcomes. Originality/value This research offeres the first empirical comparison between the new IGA interaction modes and demonstrated how integrating quantitative and qualitative objectives can support diverse design intentions across varying expertise levels.