Ooooh if deaths from societal impact is what you were talking about, then I absolutely agree with you.
The issue with AI though is that it absolutely has benefits, but there most definitely need to exist guardrails. AI capabilities are still in their infancy, AI is currently treated with some guardrails, and yet... it has shown aggressive behaviour in the pursuit of its goal. Right now you can't trust Claude to book an appointment for your at the gym without it breaking the law, exploiting software flaws, and fraudulently modifying a database it doesn't own to kick some poor unwitting person out of their reserved time slot using actions that would get a normal person at a minimum a serious fine.
AI's usefulness is not in question, it's the severity of its implications given what we already know from its limited scope.