The end result to all that wasteful training is that it doesn't work or it gets tricked or confused by junk input. Proving to be a waste of energy, time, money and the re-training, fine-tuning process has to be done repeatedly if there is an error, or everyday on new data.
Either way, it is pointless to outline a comment as 'This is a false dilemma' without countering my claims as I have both recognised the wastefulness of decentralized blockchains like Bitcoin, Bitcoin Cash, Doge, that are much worse for their intended purpose (payments) etc and using tons of data centers for Deep Learning / Machine Learning training, re-training, fine-tuning, etc everyday on user data which benefits surveillance capitalism, spyware uses etc which that is also worse for everyone.
It's pointless to counter a fallacy.
Trying to pretend machine learning and Bitcoin are of equal uselessness is dishonest.
As soon as that deep learning or machine learning model gets confused or attacked with erroneous / junk input, garbage noise or performs badly on new data, that model is useless. What do you do next? More training, re-training, fine tuning? That problem is unavoidable and evergreen.
The end result is the same and either way, that is used for surveillance, spyware on user data which not only that is a waste of energy, CO2 and time, that is worse for both for society and the planet and the methods to improve these models have been known to be inefficient and have not changed for 10+ years and always require tons of data centers. On that, Bitcoin is the same with mining. There is no benefit or useful improvement for it in the current system for payments (it's original intended purpose) and it also burns up the planet with PoW.
No idea why you had to deliberately ignore deep learning as that is also part of the planet incinerating problem.
