site banner

Friday Fun Thread for June 23, 2023

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

2
Jump in the discussion.

No email address required.

Whether they have consciousness or not is irrelevant to whether they act to achieve a certain goal. It is possible for AGI to be both non-conscious and still agentic, the same way Scramblers are.

Humans design the cognitive architecture of AGIs, and I'd imagine we would (try to) program AIs to take account of our interests. While misalignment is certainly possible, no real indication is provided in the world of Blindopraxia that the AGIs developed are routinely coming out misaligned - Captain for example seemed very well aligned with the mission it was tasked to achieve, and there's no evidence I can recall in these books of AGIs having negative influence in the larger world (if they are, they pose as much of a danger to humanity as Rorschach and Portia).

I'd imagine we would (try to) program AIs to take account of our interests

We of course would, but who says we'd succeed? Who says we'd even know whether we succeeded? We have pretty poor understanding of what comes out of where even of current models, if that continues, the future AIs would be a complete black box to us, and we'd have to pretty much rely on asking them "do you want to kill all humans today?" and trust the answer.

After all, in the same setting, people tried to control Vampires and failed. They tried to control Bicamerals and only kinda succeeded because of Vampires' help. Why would we assume they are actually in control of the AIs and not that the AIs just let them think they are, because humans tend to react violently to the perspective of the loss of control, and who wants that trouble?

if they are, they pose as much of a danger to humanity as Rorschach and Portia

Or maybe more, because they are smart enough to not reveal their intentions while there's still a chance humans can do anything about it.