WG211/M24Amin

Extrapolating from my old 2020 PL+AI graduate seminar, LLMs are good at:

• Neural-Guided Search
• Semantic Parsing: translating from natural language to formal language.

I showed experiments:

• 0. ChatGPT with formal tools
• 1. VerMCTS
• 2. Dafny-annotator led by Gabriel Poesia at Stanford
• 3. Dafny Sketcher

I am co-chairing GPCE 2025, and we will expand the scope of "generative" programming.