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.
Since the competition seems ridiculous, i checked the website and found out the competition only run for 2 days and ended at 1st April.
Are you sure it's not April Fools' Prank? or that's the way you prove that you can't predict output of psuedorandom function?
Lol, it may have been an April fools day prank. They listed a winner, so I presume they actually paid out the person/team that ranked 37th. If it was an April fools day prank, it probably had the additional purpose of getting people to learn what kinds of problems machine learning can actually solve, as Kaggle has many contests for people to solve problems that can be solved with machine learning.