What if Asimov's Laws of Robotics weren't supposed to govern robots' behavior, but the people designing them (and AIs)?

Rule 0. Do not design robots capable of harming humanity, or, by inaction, allowing humanity to come to harm.
Rule 1. Do not design robots capable of injuring human beings or, through inaction, allowing human beings to come to harm.

(more...)

@starbreaker

but zero is literally impossible as stated

I can attack someone fatally with a Roomba, and technically this is inaction on the part of the robot

if they have to all be made soft YET capable of preventing abuse, there's no reason to bother with rule 3 at all, you'd need AI running bodies made of gobs of individually disposable nanites, with tons of protection routines for humans so as to prevent inaction

but then they're nearly "indestructible" in a sense

@sydneyfalk All of that's true, but that's not the point. The point is that any machine that doesn't follow these four rules shouldn't be built.

@starbreaker

But *no* machine can follow those four rules. Any given *wheel* can't follow those rules.

People get beaten to death with crowbars, but they're 'machines' and they exist and nothing about their weaponization potential was going to stop that from happening.

@starbreaker

I mean, I can beat people to death with mechanical keyboards. How would that have been prevented?

You can't make safe hammers. If hammers can think, then you still have to decide what constitutes 'safety', and that's hazy.

I'm trying to say: There need to be explicit lines about what is ethically expected of the machine to 'prevent', because *we* can't 'prevent' our own weaponization. We cannot expect our creations to somehow be better at it than us.

@sydneyfalk I've been trying to say that it isn't the machine's job to prevent itself from being misused. It's our job to refuse when asked or commanded to make machines that we know or suspect can be easily weaponized.

We're the human beings. We should damned well know better.

@starbreaker

Well, this:

> Rule 0. Do not design robots capable of harming humanity, or, by inaction, allowing humanity to come to harm.

The last part specifically implies a robot has to prevent its own misuse -- it cannot allow, via inaction, humanity to come to harm.

So it has to *detect* misuse, so it can shut down, retract blades, etc.

> We're the human beings. We should damned well know better.

Well, based on available evidence, we don't, and never did.