Will AI replace software engineers has been the quintessential question being asked in technology circles circa 2022 and beyond. Today, I want to explore whether AI, particularly Large Language Models (LLMs), can replace software architects. The issue here is not just about AI generating code, but can it design and architect large scale complex systems end-to-end. I conducted an experiment where I gave an architectural problem to four LLMs to see if they can come up with a comprehensive blueprint of a system, like a seasoned architect would. The results were surprising, as I challenged them with designing a cryptocurrency exchange, a complex and unique domain. This experiment raises the question - are software architects irreplaceable, or are we just one prompt away from being automated?
