WShaun.com

Wallace (Shaun) Shaunfield, Copyright 2018

Posted 6/24/2018

Artificial Intelligence (AI) has got a lot of press recently. There has been a lot of activity in the field by many organizations because of the major benefits. However, many people are gravely concerned that as machines become more capable they will surpass human intelligence and will eventually overtake humans and even eliminate humans. They actually believe robots will take over and dominate humans. Is this concerned warranted? I will try to give some light on this question.

Over recent years and with an increasing rate, there have been stunning results in computer and AI technology. AI machines have beat world class experts at the game of chess, machines have been taught to learn the game, AlphaGo Zero and beat a super human version of AlphaGo, 100 games to zero. Robots have been developed to have more human like attributes. Amazon’s Echo, Alexa, and Googles, Ask Google have demonstrated uncanny results in the home such as answering questions from encyclopedias, controlling lights by voice command and many other functions. This relatively new technology of using voice to interface with the computer, provides a uniique function and I expect to see much more in the future.

Neural networks are one of the more commonly techniques used in AI machines. They consist of a black box that adjust the internal function so that a given input produces the desired output. First, the unit in training mode is given numerous input/output conditions. Then, in operation mode a new input should give the correct output. Machines using this technique have proved to be superior to doctors in diagnosing illnesses. Many AI machines use this technique. For example, the AI machine is given all the symptoms of a patient and the output is the diagnosis.

Another technique is where a guided missile is given GPS coordinates of a target. The missile then homes in on those coordinates and can strike the target from hundreds of miles away with 100% accuracy. And the list goes on and on. Is it any wonder that people are amazed at the advance in technology and are concerned as to how far this will go?

For the past 40 years, technology has advanced with uncanny accuracy following Moore’s Law, which states that the complexity of computer chips will double every 18 months (or similar time periods). I started my engineering career in 1963 designing computer chips. That was only five years after Jack Kilby invented the integrated circuit in1958. The most complex chip I designed in that first year had 40 components (transistors and resistors). Today, computer processor chips typically have 10 million components. One measure of intelligent capability is the number of computing elements. The human brain has about 100 billion neurons, many more elements than computer chips. Scientist have defined the point at which AI surpasses human intelligence as AGI, or artificial general intelligence. This is a major milestone and even more important is that intelligence will continue to increase at this accelerated rate. Some AI experts believe we will see AGI as soon as 2020, but the more common projection is that there is a 50% chance we will have AGI by 2050. Based on Moore’s Law, AI would then soon be 1000 times more intelligent than humans. We would have artificial super intelligence, ASI. There is the belief that with ASI, the new AI units would then rapidly advance the design of new advanced ASI units. You can see why some scientists are concerned about the advances in AI. But how realistic are these projections of increased technology and the resulting consequences?

Using the number of computing elements is a fairly good measure of capability if the same technology is used. Years ago, it was well known what was possible with increased elements. However, comparing intelligence of the human brain to conventional electronic technology using their respective computing elements may not be valid. Is a brain neuron equivalent to a transistor, the main electronic computing element? They both act as switches, but they still might not be equivalent. We know the brain is a massively parallel computing unit compared to conventional electronic computing, which is very sequential. It also appears that the brain circuits are three-dimensional, while our electronic circuits are two-dimensional. That would make the brain much more powerful. Therefore, the way the computing elements are arranged may be a larger factor for comparison rather than sheer number. There is talk of reverse engineering the brain, but the fact is that we do not know how the brain really works, even after many years of research. So how can you reverse something you do not understand?

There is another concept that is significant in this discussion. That is the attribute of autonomy. To have AGI, the AI unit would have to be totally autonomous; it would have to be able to think independent of any outside entity. In addition to dominating humans it would have to be mobile and be able to acquire power sources and other resources independent of human actions. So far to date, no AI unit has demonstrated the ability to think; none have come close to being autonomous. Most AI devices we see are totally dependent on predetermined conditions. Even so, we see some amazing AI capability and can expect to see more. Some devices, which can be describe as semi-autonomous, are given instructions and then released to perform some function. An example is a cruise missile discussed earlier. Once released the missile acts autonomously with one goal in mind. It is effect thinking based on predetermined conditions. However, it is a long way from being autonomous and thinking independently.

What does the future hold for creating an autonomous, thinking AI unit? I believe there is another dimension of AI that has not been discussed before. In my last book, HUMANS, I showed that the human acquisition of consciousness, which includes thinking, was a supernatural event. It clearly did not evolve, but rather just appeared in the first humans about 200,000 years ago and then was suppressed until about 10,000 years ago. It was as if consciousness was just inserted in the brain by some unknown force. This element of human history is based on evidence and strong logic and is not just conjecture.

I realize that many scientists do not believe in the supernatural, but their belief is an opinion and is not based on any evidence. They may use science and follow the evidence wherever it leads until they run into a supernatural event, at which time they switch to their subjective opinion.

There is strong evidence that supernatural events do occur. A supernatural event is one where the laws of our natural world are not followed. The actions are beyond the laws of nature. It also obvious that a supernatural event in the realm of humans would be a natural event in God’s realm. This suggest that God exists in a separate and more complex realm than that of humans. God most likely has more spatial and time dimensions than humans. I believe that God created our realm as a subset of His realm, but He did add the second law of thermodynamics which states that everything is in a state of degradation meaning our realm has a finite life. God’s and His realm is infinite –  the Bible says that God is everlasting. I also believe that occasionally when God wanted to create something in our realm the tools or capability in our laws of nature may not have been adequate. Therefore, God borrowed that capability from His realm and inserted it in our realm. Two examples of this event are the origin of life on earth and the origin of human consciousness and thinking. The origin of life was also discussed in my last book.

Furthermore, I have now come to believe that thinking itself, not just the insertion into human capability, is also a supernatural act. In support of this is the fact that the human brain may be to most powerful item in the universe. The U.S. with the new Summit computer recently captured the world’s speed record for a super computer. It can perform a particular problem in an hour that a high-performance PC would take 30 years to perform. It is the size of two tennis courts and weighs a much as a commercial jet plane. It is intended for research on AI, but it still does not have consciousness and the ability to think, at least there is no indication that it can. On the other hand, the human brain weighs a mere three pounds and while not anywhere as fast in computations, it has consciousness and can think, all seven billion that exist on earth today. Is that not an indication of the supernatural? Given this evidence it is highly unlikely that an AI unit fabricated in our realm will ever achieve the ability to think.

Assuming that an AI unit could be made to think then there are other issues to discuss. Will the AI unit have free will? How independent from society and its infrastructure will it be. Many scientists believe that such a super intelligent AI unit would be evil. What is the basis for this thought other than too many Hollywood movies? We do have an example of a super intelligent being that is very good – that is God.

Finally, there is another bit of evidence that there is no need to worry about an ASI unit taking over the world. The Bible has evidence of nuclear weapons as with the destruction of a city in a single day. The mark-of-the-beast described in the Bible book of Revelation is now possible with RFID (radio frequency identification) chips. However, as significant as AI taking over control of the earth would be, there is no mention of any such event in the Bible. It also might be that the End-of-Days prophesied in the Bible would occur before ASI were achievable.

One other area that some are concerned with is that AI will replace some humans in their jobs. As with any new advance in technology, some jobs will go away, but many more jobs will be introduced. We have seen such worry in previous introductions of new technologies. Examples include the steam engine, the automobile, automation, computers, etc.

Given all the evidence, I believe it is highly unlikely that AI will one day be a danger to humans. Rather I do see continued advances in AI that will be highly beneficial to humans. I see AI as a friend, rather than a foe.