I don't think her point disproves all possible AI doom scenarios, but nothing can. They're too vague. It's all handwaving along the lines of "a malicious superintelligence will emerge without warning and somehow seize absolute power, and it's pointless to argue about how plausible any of that is, because sufficient intelligence can do anything in ways we can't imagine." This is why the biggest boosters of AI doom conjure absurd scenarios like AI secretly manufacturing a global swarm of insta-kill nanobots. They don't care about the details, they're just supremely confident that sufficient intelligence will emerge without warning, be intractably malicious, and capable of anything by gestures vigorously methods.

AI doom critics are asking that we consider more soberly the actual probabilities of all these things happening, and on what time scale.