Overemphasizing AI autonomy

I've been noticing a trend in AI development where everyone's rushing to make their systems as autonomous as possible. Don't get me wrong, I'm all for efficiency, but I think we're overlooking the importance of human oversight. It's not like we can just set these things loose and expect them to always make the right decisions. I mean, we've all seen the disasters that can happen when an AI is left to its own devices. Take the recent example of an AI-powered trading platform that went haywire and lost millions. If a human had been monitoring the system, they could've stepped in and prevented the whole mess. But no, instead we're so focused on making AI 'smarter' that we're forgetting about the value of human judgment. I'm not saying we should ditch autonomy altogether, but can't we find a better balance between AI decision-making and human input? It just seems like common sense to me.