When developing the AIA AI Toolkit, there was considerable discussion by the AIA AI Task Force about the need to disclose the use of AI. Of course, if it is in the contract, there is nothing to debate. What are your thoughts on the topic?
Here are the highlights of what the toolkit currently includes:
Disclose substantive AI involvement:
"Disclose when AI played a substantive role in a deliverable that the client will rely on. Examples include AI-generated renderings presented as project visualizations, AI-assisted code or zoning analysis that informed a design decision, and AI-generated text incorporated into a report or specification. Disclosure does not need to be formal in every case-a clear note in a transmittal or in the deliverable itself is often sufficient."
Read the contract first:
"A growing number of owner-supplied agreements-especially federal, institutional, and large corporate contracts-include specific provisions on AI use, AI-generated content, training-data restrictions, ownership of AI-assisted work, and required disclosures. Honor those provisions. They override any general firm policy. The Contract Language tab covers what to look for."
AIA AI Toolkit: https://aifirmtoolkit.aia.org/
Related: This Fall semester, I am teaching a graduate-level seminar on AI in Architecture at NDSU. Academia is a very different context when discussing AI usage disclosure... which makes things interesting when the course is designed to specifically use AI. For each assignment, my students must submit an "AI Use & Verification Log" that includes the usage risk tier: ideation and research are low stakes; client-facing content is medium; anything touching codes, contracts, or life safety is high.
------------------------------
Daniel Stine AIA
AIA TAP Leadership, AIA AI Task Force
Lake Flato Architects
San Antonio TX
------------------------------