I disagree with both statements regardless, but there is a difference.
"There will be cases where it doesn't work" let's say that means 99%/1%. So the statements become "you would have 99% chance to be lifted out of poverty if you made good choices" and therefore "you have 1% chance of being poor if you make good choices". Whether you believe it's 1% or 0% the behavior will be the same. If a smoker gets lung cancer you will blame their smoking everytime even though there was a 5.8% chance that it was not smoking caused so 0% and 5.8% is not any different, why should 0% and 1% be?.
What you mention in your comment is logical, but you moved the goalposts by using the word "unlikely". Can and cannot are vastly different words to likely and unlikely.
Put simply, one claims it's impossible to be poor if you do X. The other claims it's possible to be poor but you have a better chance of not being poor if you do X.
For this context in reality "unlikely" and "impossible" are the same thing, barely a distinction definitely without a difference. It makes absolutely no concrete difference. But you ignored this point of my comment.
Let's say someone demonstrated that they are a xenophobe, inevitably someone will call them "racist" because it carries more rhetorical weight and is effectively the same thing (using arbitrary unrelated quality about someone to justify discrimination).
Now their "defense" will be that "no no, I am a xenophobe not a racist". How do you effectively convey that this is a bullshit defense and the connotations of racist apply equally? Without going into nuance where you will lose most people... :(