Are Biomimetic Machines Dangerous?

Posted 13 Nov 2005 at 02:36 UTC by steve

In recent years more and more robotics and AI technologies are being based on or inspired by their biological equivalents. Millions of years of evolution have led to complex safegaurds that keep those biological systems from too easily developing run-away cancers or going insane. Without similar safegaurds, could biomimetic machines development dangerous mental problems? Rodrick Wallace of the New York State Psychiatric Institute thinks so and he may have the math to prove it. He presents his ideas in a paper titled, "Entering the Blackboard Jungle: Canonical Dysfunction in Conscious Machines" (PDF format). His conclusion is that "mission-critical machines designed to emulate biological systems are likely to fail insidiously, irregularly, and progressively, particularly when necessarily operating outside their training experience".

Superiorly Inferior , posted 13 Nov 2005 at 03:29 UTC by The Swirling Brain » (Master)

Let's use inferior stuff because we're sure it will fail as opposed to superior mimmetric stuff because we're not sure when or how it will fail if at all. Like, since we can't be certain when or how it will fail, that's an unknown and that's really bad. So, let's beat our unknowns and fall back to the inferior stuff because we KNOW all about how really terrible it is! Better is not always better if you don't know it's better. In fact, let's never experiment with better stuff because if we don't we may find out it's better and then it wouldn't be an unknown anymore.

How do they know inferior stuff that we really know all the unknowns about it anyway? Like even if we knew everything, how would we really know we know everything! Therefore you just can't know everything, so you can never get a grade of 100 right!

But what if you knew you didn't know everything. That would be wise, right! Or would it?