That approach sounds similar to the Idris programming language with Type Driven Development. It starts by planning out the program structure with types and function signatures. Then the function implementation (aka holes) can be filled in after the function signatures and types are set.
I feel like this is a great approach for LLM assisted programming because things like types, function signatures, pre/post conditions, etc. give more clarity and guidance to the LLM. The more constraints that the LLM has to operate under, the less likely it is to get off track and be inconsistent.
I've taken a shot at doing some little projects for fun with this style of programming in TypeScript and it works pretty well. The programs are written in layers with the domain design, types, schema, and function contracts being figured out first (optionally with some LLM help). Then the function implementations can be figured out towards the end.
It might be fun to try Effect-TS for ADTs + contracts + compile time type validation. It seems like that locks down a lot of the details so it might be good for LLMs. It's fun to play around with different techniques and see what works!
It feel silly to say that AI is curing cancer. Normally a phrase like that would sinal the apex of the hype cycle but I guess it actually has some meat in this situation. Using AI more like statistical inference to screen for medical conditions or predict treatment could be helpful. I remember Jeremy Howard from fast.ai did that with deep learning to detect things in medical images. Seems like a good thing for CERN to do as long as it works.
Here's a fun project that you could try. Use TTS to transcribe books but make the transcriptions feel more realistic. Give each character in the book a unique voice. Leading characters should have voices based on their personalities. Use quote extraction and character attributtion to tie characters to lines. Try to do convey the human qualities with EmotionML, SSML, or some kind of semantic analysis.
The best would be a TTS system at the level of OpenAI's but with voice selection like GCP TTS so you can get quality and a range of voices.
Copyright would probably spike any monetization effort but you could try. It would be nice to have an open source tool for this though! :)
I feel like this is a great approach for LLM assisted programming because things like types, function signatures, pre/post conditions, etc. give more clarity and guidance to the LLM. The more constraints that the LLM has to operate under, the less likely it is to get off track and be inconsistent.
I've taken a shot at doing some little projects for fun with this style of programming in TypeScript and it works pretty well. The programs are written in layers with the domain design, types, schema, and function contracts being figured out first (optionally with some LLM help). Then the function implementations can be figured out towards the end.
It might be fun to try Effect-TS for ADTs + contracts + compile time type validation. It seems like that locks down a lot of the details so it might be good for LLMs. It's fun to play around with different techniques and see what works!