Three Laws of AI ? Nah

I read this article in the NYT with some bemusement. (It's an attempt to codify a more modern version of Asimov's "Three Laws of Robotics," which were his attempt to imagine how a superior artificial life form might be permanently controlled by its creators, us).

As someone who spends a certain amount of mental energy thinking about the future, and the future of technology, let me propose a truism:

The possible manifestations, both for better and for worse, in terms of human outcomes, of any new technology, will eventually emerge, regardless of any attempts to control or regulate them.

So, control genetic manipulation, even the creation of new species? Forget it. That ship is on its way out of the berth. And that means that future human evolution, way on down the road, perhaps, but inevitably, will be artificially guided. A very discomfiting thought, but count on it.

But I don't worry about AI robots taking over and supplanting us. If I'm wrong, well, they will, and RIP human race, our successors will be cyborgs. Get over it. But I don't think it will happen. And here's why.

Despite what people like Ray Kurzweil and Marvin Minsky say, I don't know of any evidence that artificial machines have ever been developed that have ANY self-awareness. Complex computational ability, yes. Modeling ability, which can look like an internally aware system, sure, but there is no suggestion that there is awareness "behind the eyes." The idea that a sufficiently complex, recursive computational system inevitably becomes self aware is actually just an assumption, a bias, even, for which, to my knowledge, there is not only scant evidence, there is NO evidence.

We really don't know what consciousness is. The problem of studying it objectively is not trivial. B. Alan Wallace and Daniel Dennett have grappled with it, but not come up with any definitive empirical view. The problem is that it's virtually impossible to examine something that is definitionally subjective and internal by using methodologies (Western scientific procedure) that are definitinally objective and external. So our only real knowledge of consciousness is experiential. We do know that artificial computing systems work very, very differently from biological ones, and that an advanced working model of a biological type brain is not even close to having been created artificially.

So, I don't worry about robots taking over. As long as we remember to always install a kill switch.

Link