I agree with the sentiment, but I also don't think such differentiation is possible, because we have no testable definition of consciousness. I genuinely look forward to the day we discover/understand the physical processes underlying our own consciousness, and how to detect/disprove it in machines, but until then... it's unfortunately unfalsifiable.

Sure, but for now, we can have a crude differentiation between an AI - a tool which has to be given an objective to work towards, and an "AC" - an entity (an intentionally vague noun) which can devise its own objectives and motivations and work toward them.