Robots rule the earth ... not!

Sep 3, 11:00 AM by Administrator

People create the funniest rumors. I know you’ve seen the movies, the one where the robot gets hit by lightning and goes haywire. Or, maybe it was made with a unique “sense of liberty” that let it break its own immutable rules. (Impossible, but still used in a story or two.) Or, perhaps you heard the one where a robot’s programming would become corrupted due to unerased experiences adding up through the years to a machine insanity: an illogical self-awareness. Regardless of what you’ve seen on the silver screen please discard it. I’ll explain the capabilities and limitations of today’s technology.

I remember when the CD first came out; and thus the world had a few CDs in place of a few hundred diskettes. Only now we’ve got DVDs packing more info than CDs, and newer DVDs that pack yet more into one disk. Not too long ago, only about eleven years ago, we hadn’t broken the 1Ghz speed barrier and I was a skeptic. Yet, here we are eleven years later with four to five gigahertz processors with as many as eight processors bundled into one. Barrier after barrier has risen to our progress, and engineers have deftly sidestepped each. But these will likely solidify into an undeniable “hard” limit to our computers in the future.

Programming limits computers far more than the hardware can. A perfect example of this is Windows® Vista: most people hate it, drivers still don’t support it, and its “improved” security is worse than Windows® XP. Don’t get me wrong, Vista is a true piece of work. But then again… how many humans does it take to overlook flaws? One! And Vista had a great many people, probably over a hundred, working all over it for years. Some bugs were caught, some security holes were plugged, and some drivers were functional, but some problems eluded everyone. Because it is such a huge storehouse of code to manage, no one person knows all the interconnections and must trust their co-workers to handle things correctly. Certainly some bugs are found cooperatively, but the ones that defy an in-depth search can be truly frightening.

Now, imagine a combat robot programmed to kill humans that don’t fit into a predefined template. Say, anyone not in a US Recon Ranger’s uniform. What could go wrong, right? Well, it could mistake a target for a tree (image recognition is absurdly difficult) or a friendly for a target; also, the robot could encounter a programming fault and freeze dead in its tracks. If this is a simple robot looking for a colored hat and with no great intelligence it would be easily outwitted, yes, but it would be much more stable in its software.

If that doesn’t reassure you, let me go one step further. The newest craze of evolutionary computing boasts programs capable of adapting to their requirements. The feasibility of putting this adaptability into robots offers a unique problem: just how fast do they learn? An adaptive robot that can learn things quickly would take years of training to meet acceptable standards, and would be obsolesced six months after release. (Though really, the new ‘bot wouldn’t be much better, just “New, and improved!”) Scientists have tried using these things in “research” without remembering that it is an emulation, and that the random numbers aren’t actually random at all.

The president of ACME Combots Inc. finally gave in to your pressure, he’ll personally order a special group of ‘bots made that learn quickly so that you can deploy them in only six months. Your new shipment of robots learn fast, work well, and are finally ready! So, you drop them on the combat field … and two days later they butcher your men. Oops! The robots learned so fast, that when the enemy changed uniforms to look like yours and kept attacking, the “friendly uniform” became the enemy’s. So the robots couldn’t tell who was friendly any more and did the logical tactic: they just butchered all the enemies (and friendlies too).

If that doesn’t encourage you, remember this: if they “lock” the learning of a robot it loses its adaptability. Humans never lose theirs, and are some of the most unpredictable and mysterious creatures to their own kind. Aside from a scorched earth campaign robots could never win the day.

A few additional reasons why robots cannot rule the earth, ever:

  1. Humans must acquiesce to let them lead, which we are far too arrogant to do. Not to mention is the fact that we already have automaton-like leaders already.
  2. Adaptable robots either learn more slowly than humans, or learn too fast to retain their knowledge. They either are locked out of learning, or are inflexible to a fault. And if they make the friend-or-foe system unchangeable, then you could just kill your enemy and put on his clothes—the robot would never attack you.
  3. Even if robots can process more than we can, robots only do what they’re told. This includes the logic we give them to determine what to do, making them very predictable and limited.
  4. Human error is exponential in group sizes, steadily increasing in severity and immutability. Because humans are fallen beings created by an omnipotent God, they are inherently limited—and just as God didn’t create another god (as much as humanists claim) so we are incapable of creating another human-level awareness.
  5. Robots cannot have self-awareness beyond the awareness of their limbs and safety.
  6. Humans are much cheaper to conscript than robots are to be programmed and assembled. Also, humans are much more apt to learn than a robot.
  7. There are two kinds of systems in a robot: hardcoded (immutable behavior), and adaptive (the learning kind). You can mix and match, but the adaptable portions are still limited by their boundaries and the unit’s experiences. There is no such thing as corrupt code that still functions.
  8. And don’t forget about maintenance, munitions, power for their batteries, communication between robots (easily jammable!), and repairs. Maintenance drones would only work for so long until they would need maintenance themselves. Robots are very complex and easily damaged, so the frequent repairs would take time.

So relax, humans rule the world and will until the end of time. Except in the case of nuclear apocalypse of course, but at least we’d all go together when we go! Robots are just too inferior, hard to maintain, hard to teach, too costly, and too easy to fool to be of any combat use in a war.