Barriers to Adopting AI-Driven Design Technologies in Architectural SMEs in Developing Countries: Evidence from a Delphi Survey in Iraq

Artificial intelligence is entering architectural design through generative modelling and data-assisted decision support. In developing countries, most of this work is still carried out by architectural small and medium-sized enterprises with limited resources and uneven digital infrastructure. Existing studies have rarely examined how barriers to AI-driven design are ordered. This paper addresses that gap in Iraq. A literature list of 30 statements was refined by a 15-member Delphi panel into 22 barriers in five dimensions. The measurement model was assessed with partial least squares structural equation modelling on 83 practitioner responses. Interpretive structural modelling and MICMAC analysis then placed the dimensions in hierarchy, and the resulting paths were tested on the same data. Economic and institutional deficiencies occupy the driving level. Technological and data readiness acts as a linking layer, while human and organisational capacity and professional risk appear as dependent outcomes. All four paths were significant, the strongest running from institutional deficiencies to technological readiness. The findings indicate that economic and institutional conditions occupy the driving level in the expert hierarchy, while human capacity and professional risk appear as dependent outcomes. All four adjacent paths implied by that hierarchy were statistically significant in the survey data.