AI Assisted Programming
Sriram K. Rajamani · 2022
We present a vision for a futuristic programming environment, where ML models and logical rules co-exist and evolve over time. We substantiate our vision using three case studies: (1) safe code generation using large language models, (2) heterogeneous data extraction, and (2) regulatory compliance for online advertising. Using these case studies, we hypothesize how such a futuristic programming system can balance productivity with safety, security and reliability, present new research results, and point to directions for future work.