An AI wolf that preferred suicide over eating sheep

Re:An AI wolf that preferred suicide over eating sheep by ncallaway on Tuesday July 06, 2021 @08:56AM (#27747866)

The issue with AI safety and unanticipated AI outcomes in general is that it’s always just a cock-up with incentives.

It’s easy to sort out in narrowly specified areas, but an extremely hard problem as the tasks become more general.

[ Parent | on HN ]

Re:An AI wolf that preferred suicide over eating sheep by qayxc on Tuesday July 06, 2021 @09:16AM (#27748087)

Even worse: if simulations are used, you now have two problems - formulating correct incentives and protecting against abusing flaws in the simulation.

[ Parent | on HN ]

Re:An AI wolf that preferred suicide over eating sheep by shrimpx on Tuesday July 06, 2021 @12:18PM (#27750563)

Isn’t this true about all systems, not just “AI”? The definition of a software bug is an unintended behavior. In a large system, myriad intents overlap and combine in unexpected ways. You might imagine a complex enough system where the confidence that a modification doesn’t introduce an unintended behavior is near zero.

[ Parent | on HN ]

Re:An AI wolf that preferred suicide over eating sheep by ncallaway on Tuesday July 06, 2021 @03:46PM (#27753357)

I think it’s true for many systems, not just AI that’s true.

AI is worth calling out in this regard because, if the field is successful enough, it can create dangerous systems that don’t behave how we want.

Building a safe general AI is much harder than building a general AI, which is why it’s worth considering AI as it’s own problem domain.

[ Parent | on HN ]