Is it possible to construct a control system where bad, fast and cheap can become good, fast, and cheap through repeated sampling and a strong spec/eval harness?
I am trying to keep an open mind with AI, but I also have little understanding of control theory, trying to learn.
You can, but you need to break the problem into much smaller tasks, then check those answers, and finally have a harness that handles all the context, task breakup, task definitions, and validations each round.
If the model provides me with bad results because it's dumb, I don't care how quickly it does it.
But there are lots of use cases where a relatively "dumb" model is good enough.
Is it possible to construct a control system where bad, fast and cheap can become good, fast, and cheap through repeated sampling and a strong spec/eval harness?
I am trying to keep an open mind with AI, but I also have little understanding of control theory, trying to learn.
You can, but you need to break the problem into much smaller tasks, then check those answers, and finally have a harness that handles all the context, task breakup, task definitions, and validations each round.
Smaller models seems to get stuck in "loops" when you try to "handle" them this way.
fast results that you need to verify are better than slow (allegedly better) results that you still need to verify. REPL vs batch.