Is AI a Real Threat to the Human Race?

 It is a fact that Artificial Intelligence scares the crap out of some people.  The knowledge and
resources are in place that allow for the construction of more powerful AI systems.  It doesn't stretch the imagination that  such AI systems could reach the level of humans in intelligence and in doing so have the capability to construct AIs that could be far more intelligent than the original AI.  

Present AIs are fairly cute and simplistic.  However as the past has demonstrated, as the systems become more capable, more responsibilities will be shifted to them.  The question on a lot of people's minds are how long before we can no longer control these computer intelligences?

As the author postulates in the Irish Times Technology section:

• Imagine how a medical robot, originally programmed to treat cancer, could conclude that the best way to obliterate cancer is to exterminate humans who are genetically prone to the disease.

People who support AI postulate that things like this would never happen - that programmers will  build in safeguards.  This, author Nick Bilton says in part, from programmers who can barely keep present day computers from crashing whenever you check just your e-mail

From the the Irish Times  article:

• Elon Musk, recently said artificial intelligence is “potentially more dangerous than nukes”. Stephen Hawking, one of the smartest people on earth, wrote that successful AI “would be the biggest event in human history. Unfortunately, it might also be the last.

Computer That Thinks Like a Person?

Are you familiar with the Turing Test? The Turing The Turing is measure of artificial intelligence. It is arrived at if a human can have a conversation with a machine and not know they are talking to a machine. If so the computer or more to the point, it's programming is thought to be as 'intelligent' as a human. In most cases however the programs simply mimic how a person reacts and is easily caught when asked to perform outside it's program parameters.

Now there is a program called "CLARION", a program that performs the same way human subjects does in cognitive tests. It provides results that are not so much what we think, as earlier programs did, but how we think. And that is a huge difference.

When programmed, Clarion was not filled with pat human responses but a set of rules instead that were modeled on how human subjects went about solving a problem.

From the IO9 article on how the results were formulated:

• When humans were asked to discuss their thoughts about the problem prior to giving a solution, they answered correctly 35.6 percent of the time. When they were assigned another task, interrupting their work on the problem before going back to thinking about it, they answered correctly 45.8 percent of the time.

This is what happened when Clarion was put to the same conditions:

• When CLARION solved problems under similar conditions, the results uncannily mirrored those of human subjects. Five thousand runs of the program returned a correct result 35.3 percent of the time when the scientists imposed conditions similar to a human 'discussing' their work, and 45.3 percent of a time when it was interrupted to work on another task.

Does this mean that the program is thinking just like a person? Well, not really, but it does show some of the possibilities and what we can expect from expert systems in the future.

Wiki Turing Test article complete IO9 article Wiki article on Cognitive Architecture

Creating Artificial Personalities

The Daily Galaxy has posted a great article written by Luke McKinney, concerning ongoing work to produce artificial personalities. Now before we get all hyper weird on this subject, I have to say that I am a believer in the split between mind and body, or the brain. The brain is the structure capable of supporting the software that we think of as the "mind". The reason I am bringing this up at all is that at some point researchers are going to be able to build hardware that will be on par with a human brain. But just building the hardware does not imbue it with a mind. But hardware capable of the massively parallel processing that the human mind is capable of requires software well beyond the if this happens do this, that is prevalent in today's robotic equipment. If we want to get anywhere near Asimov's positronic powered robots, we are going to have to throw out virtually all of the presently held ideas on how a robotic or artificial personality would operate.

But even so, A research collaboration between Samsung and the Korea Advanced Institute of Science and Technology are tackling this enormous project in an unusual way and with an unusual model.

The article is something that should be read in it's entirety here

Scientists create LIVING artificial nerve networks

"They're made out of meat!"

No, (with all due apologies to Terry Bisson), not the humans. Their computers. Soon.

Because in the future, the interface between brain and artificial system will be based on nerve cells grown for that purpose. As recently featured on the cover of Nature Physics, Prof. Elisha Moses of the Physics of Complex Systems Department at the Weizmann Institute of Science in Rehovot, Israel, and his former research students Drs. Ofer Feinerman and Assaf Rotem, have taken the first step in this direction by creating circuits and logic gates made of live nerves grown in the lab.

Here's a summary what they did, according to their published results "Reliable neuronal logic devices from patterned hippocampal culture" Nat Phys 4, 12 (Dec 2008) Ofer Feinerman, Assaf Rotem, and Elisha Mose

The researchers grew a model one-dimensional nerve network – by getting the neurons to grow along a groove etched in a glass plate. They found they could stimulate these nerve cells using a magnetic field (as opposed to other systems of lab-grown neurons that only react to electricity), and that varying the width of the neuron stripe affected how well it would send signals.

The scientists took two thin stripes of around 100 axons each and created a logic gate similar to one in an electronic computer. Both of these “wires” were connected to a small number of nerve cells. When the cells received a signal along just one of the “wires,” the outcome was uncertain; but a signal sent along both “wires” simultaneously was assured of a response. This type of structure is known as an AND gate.

The next structure the team created was slightly more complex: Triangles fashioned from the neuron stripes were lined up in a row, point to rib, in a way that forced the axons to develop and send signals in one direction only. Several of these segmented shapes were then attached together in a loop to create a closed circuit. The regular relay of nerve signals around the circuit turned it into a sort of biological clock or pacemaker. Full article here

Image source: Nature Physics magazine.

Electronic Sherlock Holmes under development

Neo-Moriarties look out: the University of Washington will lead a multi-institutional group pushing the limits of computers' ability to interpret data and ultimately predict the behavior of complex systems. "A complex monitoring system has far too many pieces of information for any one person to look at," said principal investigator Pedro Domingos, a UW associate professor of computer science and engineering.

The basic approach is the same as that of fictional detective Sherlock Holmes: using the powers of reasoning to discover the best explanation for a set of clues. But today's reasoning can't be done by a single, pipe-smoking sleuth. The modern military has millions of possible clues, including sensors on soldiers, satellite maps, road monitors, aerial drones and written observations from reconnaissance missions. The Army Research Office that provided the grant wants to make sense of this information in order to make decisions and predict an adversary's next moves. About the designer