The research on engineering software applications that employ artificial intelligence (AI) and machine learning (ML) is at an all-time peak. However, most of the research in this area is focused on the interactions between humans and AI which, in turn, is predominantly concerned with either building immersive interfaces and user experiences that allow for increased telemetry or on handling AI and ML applications in production (MLOps). Nonetheless, the research on fundamental architectural differences between AI-powered applications and traditional ones did not receive its fair share of attention. To that end, we believe that a new take on the fundamental architecture of building software applications is needed. With the ever increasing prominence of content-driven AI-powered applications, it is our conviction that 1) content could be served by servers without clients requesting, 2) servers could (should) request data from clients without waiting for their requests, and 3) interfaces should dynamically adapt to updates that happen to the intelligence driving the application. Hence, in this paper, we propose the fluid architecture that facilitates the bidirectional interaction between clients and servers as well as accommodates the co-dependent evolution of interfaces and back-end intelligence in AI-powered systems.

@inproceedings{paperbot6558,
  title = {Towards Fluid Software Architectures: Bidirectional Human-AI Interaction},
  author = {Ammar Yasser and Mervat Abu-Elkhier},
  booktitle = {ASE},
  year = {2021},
  doi = {10.1109/ase51524.2021.9678647}
}