Is Artificial Intelligence the Next Step for Control Systems?

The current limits are pretty much a zero as we really have only scratched the surface of the technology and have nothing deployed.

As far as the future goes, that's hard to predict, but there really are no limits except for the imagination of the creators.

From the linked article, "A fundamental problem, though, is that nobody appears to know what intelligence is."

Well, I am coming more to the belief that the definition is less important than is being proposed. If the problem space was reassigned to attributes or functionality that are necessary to meet a specific set of requirements, then there is no issue with the definition, just the means to execute the design.

Nor do I feel that a general consensus is required for what AI can be used for. That's just as silly as inventing a pencil and demanding what can be written with it.

I think the root problem is that we do not, as of yet, know how to achieve or replicate the cognitive decision making abilities of the human brain in a synthetic sense. The closest we have come to doing that is the old fashion way of making babies.

That doesn't mean we won't be able to get there at some point, just that we are only at the cusp of making inroads in that direction.

Lastly, AI gets a lot of attention these days as it is a hot button in pop culture. AI is a pretty dull concept in of itself. All it means is that a synthetic system can mimic certain decision making processes that the human brain does.

That's a long way away from an artificial or synthetic consciousness and possibly further still from true sentience based on a non-biological platform. Yet, pop culture likes to lump all of these things together, partly due to the entertainment industry's monopoly on our imaginations.

AI will be nothing more than then the first paving stone in a long road of a progressive endeavor of building more and more clever machines that can make valid decisions and extrapolations based on available data.

These systems will find their way into manufacturing as well as every other aspect of daily life, just like the integrated circuit does today, given enough time and imagination.

Perhaps someday, AI will fill the gaping intelligence void in Congress and our state capitols.

There's nothing there now.

Never happen, Lyn. AI requires rules to be followed and they don't do that!

Well I think definitions are important from a legal standpoint, there must be defined uses of AI, and so follows, what constitutes AI....Such as we would not want autonomous control of an AI soldier that includes deadly force...life and death situations are all around us, we must carefully set the parameters of what can and should be controlled, and by who or what...and a definite path of responsibility....The last thing we want is a wrongful death claim where there is no accountability defined....however that is what it might take to set precedent...We have enough problems now trying to define who is responsible for what and why.....as AI becomes more and more human like in it's abilities, so too must the parameters of responsibility and conditions...for instance a machine kills a human, it's found that the software was infected with a virus of unknown origin....who takes responsibility? The builder of the machine, the designer, the software engineer, the operator?? One only has to read a few disclosure agreements to see how carefully they are worded to hold nobody responsible....and what about autonomous thinking robots of the future, that make decisions independently,,,,what will be their legal status?? Will the time come when an individual robot can be held accountable for wrongdoing? Yes I think so.....so yes the level of intellect and the level of autonomy will come into play....and we must prepare....

"who takes responsibility?"

The first is the one with the deepest pocket, then it follows in descending order.

"who takes responsibility?"The judge will decide, who is also a robot!

Yes but there needs to be some ground work laid, for risk assessment purposes...To move a technology forward takes venture capital, which requires risk assessment, which includes liability issues....

I agree, responsibility is highly important.

As machines get smarter (AI) even if we humans decide the machines can be held accountable, we are faced with a dilemma.

As more AI machines become available, the price will drop. Maybe not to an affordable level by the average person's standard, but to a point where one may not have to be a multimillionaire to afford one. And they will become affordable by thousands of businesses.

Then those with the means will be able to use the machines to do dishonest things, even murder. The machine will be held accountable and destroyed....or reprogrammed. The evil deed has been done, society is under the false impression that justice has been served, meanwhile the evil doer is planning the next deed.

Heck, someone may even make a business out of it with the slogan "Dirty deeds, done dirt cheap".

That's no different than hiring a hit-man. You are still culpable as is the assassin.

Nothing new here.

That may be true, but I suspect it will be easier to hide the trail leading back to the originator. A hit man, if caught red handed, may be more likely to give up his employer than a programmed AI machine.

I like the pic, but he's only part machine, and his "I" is not artificial but natural.

I would think there would be far easier ways to do someone in than to trick an intelligent machine into doing it for you. I would be more worried that the intelligent machine would decide to do it on its own!

Here is a paper on applying neural networks to control systems.

http://www.cs.dartmouth.edu/~spl/Academic/AI/support/Neuralnetworksforself-learningcontrol.pdf

Just to provide some entertainment on the subject, a very good friend's late father created a maze solving computer in 1952 using relays and solenoids. While the technology is stone aged compared to modern computers, it demonstrates some elementary principles in AI problem solving.

I was given one of his original manuscripts for this device that he built for the Franklin Institute. I find it really amazing what they were able to do 60 years ago, in what Spock would call a stone aged zinc-plated, vacuum-tubed culture, and how it set the stage for modern AI development.

Modern folk aren't any smarter, they just have better tools and higher shoulder to stand on.