[abalashov]

I would be extremely wary of making such cosmological, sweeping generalisations about anything. "Realistic hope that the person will ever be non-harmful" really depends on a gazillion other variables.

And why start with a semantically negative presumption, anyway? A world of difference in attitude can be reaped from daring to ask whether a person might harm again, rather than if they will "ever be non-harmful".