Well checked the PDF and author out a bit more, the guys seems legit and real. The paper seems to be a summary of an approach taken that they thought might give some statistical edge using the Deep Learning AI system.
Outcome = epic fail.
The outcome was only one of the numbers selected dropped.
So a new question arises, is AI fallible?
I have done some superficial NN (Neural network) predictions using various algorithms that I simply "stole" off the web and programmed into excel. Can't say I was impressed with the outcomes, essentially it tries to predict a future outcome based on past outcome predictions and fails that have then had a "real" correction applied to them. (My very limited description of them).
So you essentially end up with a guess based on a guess that was wrong and corrected using the actual outcomes and then using the corrections for the past wrong guesses that where corrected on the fly in an effort to reduce the error over time to ensure a closer guess to the outcome you are trying to guess, or something kinda like that.
So for me the simple answer is yes AI or NN is fallible, so when you hear AI robot Sophie and her counter parts discussing the downfall of humankind get ready to turn the power off.
As for lotto results derived from theses approaches I read something that I can no longer find where a study was done comparing the outcomes of lottery predictions using AI, NN and other current statistical approaches including excel based forecast and trend functions.
Turned out that the "overall" results when applied to lottery and also market trending was essentially the same. They all came within a specific performance range and accuracy with some, and even the lesser sophisticated performing better in specific categories.
A name that comes to mind is Warren Buffet, a mastermind of the markets and predicting their performances. If you check him out he is always on about the behaviour in things, he has understood that a dynamic performing system is subject to behavioural inputs that change the outcomes. He seems very adept at picking or choosing what behavioural trends are having an effect in specific market areas. Hence he's a rich old boy.
I know next to nothing about markets? Clueless.
But I do see parallels to the behavioural patterns I have noticed in the bunnies of the lotto.
So moral to the story?? Don't rush out and buy the latest lottery number prediction software based on AI or NN algorithms that "will" give you that winning edge.