“LLMs on the boundary | deadSimpleTech”
https://deadsimpletech.com/blog/llms-on-the-boundary
> In general, getting something halfway acceptable out of an LLM means running what is essentially a Monte Carlo simulation: you prompt the model over and over again, changing the prompt slightly each time, until you hit something that meets your acceptance criteria.
The article claims that LLMs have advanced in software development, due to the ability to have tests written to verify the output; this assumes a test-first development method. The main issue with test-first development is wrong assumptions. The article somewhat covers this with the explanation of the impossibility of bounding natural language, but its not explicit.
Bounding a software space was already the biggest issue in software development. LLMs just add to the problem.
