SAG Expert Interview #03: Impact of AI in Software Architecture
summary
Reports experiments over two years using prompt engineering to get LLMs to offer trade-offs and to ask clarifying questions, but finds LLMs excel at rhetorical tasks and fall short at abstract system modeling. Argues the architect stays accountable — AI gives suggestions, humans decide — and admits making one mistake a day while resolving them quickly. Predicts either further acceleration toward more humanlike multimodal models or a plateau where AI becomes commonplace. Emphasizes building conceptual integrity across domain, software, and organization.
AI-written summary · about
chapters (8)
- 0:00Introduction and background
- 1:02AI in software architecture
- 2:50Accountability and AI limitations
- 3:54Architect's daily challenges
- 5:57Complexity in architecture
- 7:22Future of AI-human relationship
- 9:31Israeli culture and innovation
- 10:56Value of conferences and networking
related talks
50mAI: A Tool for Software Architecture?
21mAI's impact on software architecture
45mThe AI Software Architect: Revolution or Illusion?
09mLessons Learned in AI Engineering
27mThe sociotechnical debt of AI breaking Conway's Third Law
59mIt AI n't What You Think!
22mAdapting to the Future of Work in Tech
07mAI's impact on programming jobs and developer skills