Now since Quantum works by learning mechanism that is you give teach it some examples and based on which it can intellectually learn billions of inputs and outputs based on that data.
So one day it could happen that there will some meaning to these seeds based on mathematical calculations. It will need huge computing power may be 100's of teraflops at any given time but it could be possible to know what was done on that seed.
I hate to break it to you but you're mixing up machine learning with quantum computing. Those are two wholly different technologies based on fundamentally different concepts.
To make matters worse, not even machine learning works as described above. Cryptographic hashes by definition output pseudorandom data. There's no pattern or bias to be found and accordingly nothing for a neural network to be learnt.
Kaggle ran a
competition several years ago in which competitors had to create a model to predict the output of a pseudo-random generator. The prize was $1000 for the 37th place model (divided by the number of entries the team placed), and there were no other prizes.
I am not quite sure what point the OP is trying to get across. Provided a working RNG is used to generate a seed, you cannot calculate a SHA-256 private key based on a public key, or based on an encrypted message to the private key with current, non-QC technology. In theory, QC will be able to crack some types of encryption in the future, but this is still theoretical.