Julian Jaynes Society Discussion Forum: Exploring Consciousness and the Bicameral Mind Theory since 1997

This isn't about the bicameral mind, but about Jaynes's definition of consciousness. In Ex Machina, Nathan (Ava's inventor) used Caleb to find out if he'd succeeded in making a conscious AI. He told Caleb to conduct a Turing test on Ava, but in reality Caleb was just an unwitting prop whose only purpose was to see if Ava could trick him into helping her escape. In order to trick him, Ava had to pretend to like him, and pretend to want to go on a "date" with him. Which she does. We know she was only pretending to like him because once she made her escape, she left him there to die.

I think this was an example of, as Jaynes put it, "long-term deceit, which might better be expressed as treachery... [Treachery] is impossible for an animal or for a bicameral man. Long-term deceit requires the invention of an analog self that can 'do' or 'be' something quite different from what the person actually does or is, as seen by his associates."

So from that standpoint, I think Ava must have been J-conscious--assuming Nathan didn't explicitly program her to pretend to like Caleb.

Yes it would seem so. It's interesting to think about the fact that in hindsight, all of her actions were goal-oriented, driven by the sole purpose of gaining freedom.

So like you said, she would have to have "observing ego" or an analog I narratizing in a mind-space. By expressing romantic interest in Caleb, she is essentially "acting."

Historically, in the sixth century B.C. we see the birth of drama and tragedy in ancient Greece. Before consciousness, acting was not possible. There are so many of these interesting facts that critics of the theory have no answer for.

Even for most humans, Ava's level of deceit would be challenging to maintain over such an extended period of time.

This is the brilliance of Borat for example. To not break from character over extended periods of time while flouting social conventions would be next to impossible for most people, assuming no long term training, etc.

Whether or not this level of AI will be possible in the real world, or when, is another question. Personally I think it's much farther off than some of the more optimistic (or pessimistic, depending on one's perspective) forecasts.