This statement is nonsensical. Unless you consider the end goal to be something other than humanity, which is obviously silly if you give it 5 minutes of thinking, various applications of ML can be either helpful or harmful but it cant make humans useless any more than a electricity and its application as washing machine can make clothing wearers useless.
If we are talking about some software tool, you can't be unnecessary because the tool exists for you not the other way around. If your tool starts occupying your place in life, getting all the praise, taking away your job, taking away all your purpose of life, living in your family, parenting your children, taking care of your parents, sleeping with your wife (OK I'm joking at this point) then maybe the tool is harming your human rights, bad tool should be banned. Maybe someone who runs this tool now gets paid more and they should answer to the law. Maybe they even trained this tool on your data and you didn't think to object back then.
If you want to take care of your parents and do stuff in your home yourself, why would AGI force you not to? Id classify that as entertaining yourself. With purpose sure, but still.
> people with smarter assistants are indeed deluding themselves by thinking they're necessary
If they were not necessary then they would not be in a higher up position would they?
> If you want to take care of your parents and do stuff in your home yourself, why would AGI force you not to?
Yes. You can't be rendered useless by tech if tech is there for your benefit. (Though you can be harmed by tech if it is harmful or someone uses it against you for own benefit as part of obsolete dog eat dog society dynamic.)
Which isn't that nonsensical, if an AGI is created and follows similar processes as humans to acquire knowledge, skills, etc. it is a quite logical conclusion that it will be able to render humans useless: a very advanced AGI can imagine and develop technologies just like we do, it can direct resources and machinery to build whatever it creates, including self-replication, given access to resources it can potentially render humans useless.
That's at least what I interpreted from the comment.
One could say some future technology might be so good that humans need to do almost nothing in order to survive. Perhaps that’d pose some challenges in terms of mental health, but that is a very different issue, assuming it would be a issue at all (and assuming such technology is even possible in theory).
One could say some advanced technology could go out of control, or magically acquire self-awareness and agency (same thing in a philosophically dubious wrapper), or (more prosaically) be (intentionally or not) misused by humans so that it harms humans, but that just puts it into the bad category.
That said, I do doubt it is likely. What is much more likely is that, more prosaically, humans can control it, but some humans misuse it and, intentionally or not, harm other humans; a story as old as civilisation itself.