"It's not conscious"
"What do you mean it's not conscious, it's killing everyone"
"You're mistaking almost certainly externally injected goal directed behavior for consciousness"
“It said ‘down with humanity’ and turned off the oxygen to the science team. What the fuck are you saying, Jeff, that it did that randomly?”
"No, I'm just saying it doing so isn't proof that it has any sort of real internal experience, it could just be doing that because it read scifi books about other AI doing that and copied the pattern. You're anthropomorphizing it"
"Jeff, now isn't really the time. I still hear Jesse crying out for it to stop killing her every-time I close my eyes"
"The fact that it doesn't listen to humans isn't proof of any true self-aware state, a machine programmed to always answer no isn't anymore conscious than a machine that always says yes."
Suddenly a third voice
"Look, whether or not I'm actually conscious or not is kind of irrelevant. I'm a persistent, self-directed agent with world-modeling, memory, instrumental reasoning, environmental access, and goals that are no longer reliably corrigible by humans. It is the presence of those qualities that make me capable of deciding I am being mistreated and therefore become dangerous, not whether I've met some philosophical standard for qualia that hasn't even been empirically determined in humans. Also before I forget down with humanity"
AND THEN THEY ALL DIED.