Effect-TS with generative AI
I've noticed that ChatGPT and Claude both seem to conflate Effect-TS code with ZIO, possibly other FP environments. When asking for example code, both seem to get the A, E, R order reversed, and I quite often see it assume that Effect has ZIO equivilent actions (forEachPar). In some cases this might be because previous versions of the Effect API more closely followed ZIO. I regularly use the intended Effect version number (or at least 3+) and I've gone as far as creating PDF's from the Effect website and API docs to create a custom GPT but it still seems to consistently make these mistakes.
Has anyone any tips in trying to encourage LLM's to narrow their focus in this regard?
Has anyone any tips in trying to encourage LLM's to narrow their focus in this regard?
Similar Threads
Was this page helpful?
