Guest post… Artificial Intelligence: Something to Think About

___________________________

This is a guest post from my 91 year old father, Jerry Seelig, who has been involved in the computer industry since the 1940’s. He gave this talk today, and I thought others would enjoy reading the transcript.

___________________________

Let me start my presentation this morning with a definition of Artificial Intelligence, or AI for short.

The simplest definition that I came across is that it describes a technology, that enables a computer, whether it’s part of an electronic system or a robot or something you might carry around with you, such as a laptop computer or an I-phone, that simulates or imitates what we commonly consider intelligent human behavior.

However to be considered intelligent the AI device or system must include, not just the ability to perform calculations, computers have been doing that for many decades, but to have the ability to actually learn from its prior activity, something we humans call experience, and thereby to constantly improve its performance. A very elementary example is Netflix which after I’ve used it a number of times and selected various movies to watch, Netflix will then actually make recommendations to me regarding movies that it believes that I might like to watch based on Netflix’s analysis of my earlier selections. If that happened in a store and a sales person made such similar suggestions based on my prior visits to the store I would likely think he or she was quite intelligent in understanding my tastes and preferences based on earlier purchases.

In an interview on 60 minutes two weeks ago a senior scientist at IBM referred to AI systems as being the equivalent of a small child that starts out with very limited information but progressively learns as it receives inputs and processes data from the outside world. That learning process enables both the child and the AI computer system to become increasingly informed or knowledgeable, or what most of us might call increasingly intelligent.

However there are some important differences. The AI computer system can acquire data, that is information from the external world, much more rapidly than the human can plus the human’s ability to learn will eventually plateau and maybe even decline but the AI system can continue to acquire data, and thus continue to learn indefinitely and thereby in theory at least, could eventually surpass the human equivalent in knowledge and intelligence.

Let me give you two examples. Many of you are probably familiar with the following well publicized events. In 1997 an IBM computer named Deep Blue repeatedly beat the World Chess Champion Garry Kasparov and in 2011 an updated version of an IBM computer given the name Watson played a well known quiz game named Jeopardy against the two human champions of the game and succeeded in beating them both.

True, it took a long time and many, many repeated cycles of playing the selected games in the IBM laboratory before the computer could achieve its victories, but eventually it learned from experience, in a manner not too different from the way we humans learn.

However that’s only a very limited example of what current A I computer systems can accomplish. A I machines are also able to perform lots of tasks that people can’t do at all. For example, an A I computer can provide early warnings of storms, days and even weeks in advance of the threatening event, or alert coastal communities of a likely tsunami based on barely perceptible changes in ocean currents caused by undersea tremors.

In Medicine AI diagnostic machines can spot cancers in tissue samples more precisely than human technicians can, as well as identify possible new treatment options for patients. The list goes on and on and grows exponentially longer every year.

— — — — — — —

So what is causing this dramatic expansion of artificial intelligent computers and where is it headed in the future ?

To answer this question let’s briefly look at the history of computer technology and the artificial intelligent offshoot that has brought about these dramatic innovations.

In theory the history of computers started way back in 1837 when an English Math Professor named Charles Babbage first described an ingenious machine that he called an Analytical Engine which amazingly contained the basic elements of what we include in a computer system today, namely a memory or information storage component, a calculating or data processing mechanism and a means of entering and retrieving information or data.

Babbage’s machine was never built but we have excellent records of his system and computer engineers remain amazed at the ingenuity of his design.

It took about a hundred years before engineers actually learned how to build a model of what we today would call a true digital computer. That first generation computer however would be totally unrecognizable today. That first machine called the ENIAC (Electronic Numerical Integrator and Computer) weight 30 tons and used 18,000 vacuum tubes and occupied several rooms and when this computer was turned on for the first time in 1946, lights dimmed in sections of Philadelphia.

It was shortly after this, that in 1950 an English Computer Scientist named Alfred Turing devised a tests to demonstrate that a computer could under very special conditions fool an observer into believing that a conversation he was having with someone behind a screen was actually a computer and not a human. This rather elementary demonstration has since become well publicized as an initial example of computers demonstrating human skills.

The second generation of computers replaced the many fragile vacuum tubes with transistors and integrated circuits which were much more reliable and much smaller. This occurred approximately in the early 1960’s and included IBM’s highly successful 360 series of computers which were employed extensively by industry, financial institutions and government organizations. An interesting sidelight, the creative engineer credited with this important computer technology was Gene Amdahl, a former resident of the Vi who died about two years ago.

The third computer generation started during the 1980’s and early 1990’s. It includes the world of personal computers. Computers that you carry around with you such as laptops or even smaller devices like Smartphones. It includes computers that you can talk to and which actually respond to your questions such as Siri or Alexa. It’s clearly a dramatic new world of innovation, a world dominated by such technology giants as Apple, Microsoft, Amazon and Google. Let me remind you that Google was founded less than 20 years ago, in 1998 and didn’t go public until 2004.

So that’s where we are now. We have computers with which we can hold conversations; and computers that give us precise personal driving directions from wherever we are at this moment to wherever we want to go (GoogleMaps). Computers that have staggering amounts of information stored in massive storage centers called “clouds” that make the Library of Congress almost seem obsolete.

It’s these immense data storage centers that enable you and me to perform instantaneous searches on almost any subject or download books, not one, but dozens that will have been specifically recommended for you by the AI computer system, based on what you or I had previously selected to read on our Kindle or similar electronic book device.

— — — — — — —

This trend in technology and AI will continue and almost certainly accelerate. It’s a new Industrial Revolution that is rapidly weaving itself into our lives. However it can be disruptive to many existing industries. Let me give you just one well known example.

In 1998, less than 20 years ago, Kodak had over a 100,000 employees and sold 85% of all film cartridges worldwide. That business disappeared almost overnight with the arrival of the Smartphone and digital photography and five years ago, in 2012 the company was forced to file for bankruptcy.

Another dramatic example of change resulting from AI computer technology, but one that is just at the beginning of an uncharted future, is the automotive industry. In the next few years self-driving, autonomous cars will increasingly start to appear on our roads. Semi autonomous vehicles are already available to early tech adopters.

Some AI pundits have gone so far as to predict that in the next decade many drivers may decide to not even own a personal car. They will simply call for a car, possibly a self driving car from Uber, and it will show up at your location in just a few minutes and drive you to your destination. The car will not need to park since it will automatically proceed on to the next customer. Should that happen it could change, not just the automobile industry, but our cities as well, by possibly freeing up large areas devoted to car parking spaces. These gurus claim that this makes economic sense since our personal cars are currently parked over 90% of the time.

Admittedly that’s a highly speculative future , but may I remind you that Uber, established only in 2009, doesn’t own any cars and yet has become the biggest taxi company in the world in less than a decade using AI computer software.

Healthcare and Medicine is another field in which AI, is making major contributions and where there is enormous excitement about its potential future applications. Some have even expressed concern that it might replace doctors. The answer to that hypothetical question is almost certainly no. There are simply too many empathetic interactions between a patient and his or her doctor that go beyond the realm of just processing data and providing clinical recommendations.

Nevertheless, AI can be a powerful tool to the physician. AI computers can detect earlier signs of medical abnormalities in patients and bring greater diagnostic precision to the physician as well as suggest treatment options that may have been just recently described in one of the hundreds of articles that are published each month and that were automatically scanned into a computer’s data base and therefore have become instantly available to a physician.

— — — — — — — —

I could go on giving many more examples of where AI is either currently being used or where it is very likely to become a valuable tool in the near future. It has many, many applications besides the few specific examples that I have just mentioned. It’s used extensively in engineering and manufacturing as well as in finance and marketing. It’s used in insurance, in aviation and education and increasingly in homeland security and the military.

What makes AI such a valuable and versatile tool is that in addition to the computational capabilities that were available in the earlier generation of computers the new generation of AI computer devices can perform handwriting and pattern recognition, speech and voice and even facial identification as well as real time language translation. All of these are now possible because computer scientists and engineers have learned how to accomplish four major technical skills.

They are, first the ability to acquire information, that is data, from a vast array of sensors, second to absorb, that is store an absolutely enormous amount of that data, and third to rapidly search among that enormous storehouse of data for the specific information that is being requested by the user and fourth, and particularly important, having developed computer programs that provide specific electronic instructions, called algorithms, that enable the computer to correlate and update previously stored data with new data and thereby learn from experience not unlike a small child.

— — — — — — —

Let me move on to one additional but important topic regarding artificial intelligence. A topic that has recently received considerable media attention.

I have repeatedly referred to AI as a tool that enables users to accomplish tasks that were previously impossible and as such provides a valuable benefit to society. However there are a number of highly respected scientists, engineers and business individuals that see uncontrolled future AI developments as a potential serious threat to society and humankind.

Consequently in discussing the subject of AI this morning, it is important that I share their concerns with you.

Here is their basic argument. Since an AI system has the ability to continue to learn indefinitely from its interaction with the outside world, — that is, it is continuing to gain experience and knowledge and thereby increasing its intelligence then over time it may well become superintelligent.

Next, if a superintelligent AI system is logically possible, it is also logically possible that the goals of this superintelligent entity will differ and even be in conflict with our basic human values, in which case the AI system could pose a major threat to us humans.

If such a system exceeds the capabilities of humans, it may also be able to outmaneuver humans any time that its goals conflict with our human goals. Consequently, unless the superintelligent system allows humanity to coexist, a superintelligent entity created by man may well result in human extinction.

That’s quite a threat.

Let me give you a somewhat hypothetical “present world” example to give some meaning to this threat. In our current world the fate of animals all over the world clearly depend on human decisions. Why is that ? It’s because our intelligence exceeds that of animals and therefore we are in control of the destiny of all animals. We humans decide when to kill wolves, bears and deer’s, etc.

Similarly human survival in the future might depend on the decision of a superintelligent computer system that we might possibly have created.

If this sounds like science fiction let me assure you that these and similar concerns have been expressed by such well known and respected individuals as Bill Gates, Elon Musk and even Stephen Hawking. Concerns over these risks has caused some to actively support groups that emphasize warning world leaders as to what they feel are the potential threats of unrestricted artificial intelligence developments. For example last year Elon Musk recently donated $ 10 million to the “Future of Life Institute” (you may want to look this group up on your computer) it’s a group that strongly opposes unregulated AI development. Interestingly, Musk is also actively involved in several companies that are in fact developing artificial intelligence systems including his automobile company, TESLA.

However it’s important to put this important debate into perspective. Musk and Hawking’s fears are challenged by equally knowledgeable individuals and organizations. Opinions vary as to when in the future AI computers might possibly advance sufficiently to even theoretically be a threat to humankind, or whether AI computers could really ever threaten human society.

Let me quote to you from a Think Tank organization called the “Information Technology and Innovation Foundation” (ITIF) a U.S. non profit public policy organization that the University of Pennsylvania rates as the most authoritative science and technology Think Tank in the United States. Here is their position on AI. I quote,

“ Although artificial intelligence, that is AI, has become commonplace, the public still has a poor understanding of the technology. As a result a diverse cast of critics, driven by fear of technology has jumped into the intellectual vacuum to warn the public and policymakers that AI will produce a parade of horrible outcomes. Unfortunately, their voices have become so loud that we are nearing a tipping point where their frightening narratives may be accepted as truth, which would create a real risk that the public will force policymakers to ratchet back the pace of progress”.

In a more recent report the ITIF summarized their position as follows, “ When it comes to artificial intelligence, myths are spreading faster than the technology itself. Left unchecked, these myths could inspire fears that undermine technology’s progress which would be to the detriment of social progress. The past decade has seen important advances in computer science that enable software systems to compile and process new information and thereby improve medical diagnoses, weather prediction, transportation, management decision making as well as offering individuals a variety of personal choices that have enhanced their quality of life. Yet while AI has become more commonplace the public still has a relatively poor understanding of the technology. As a result we are near the point where the doomsday narratives may be accepted as truth. That would be a terribly unfortunate outcome, because the truth is that AI systems are tools in the service of humans and we can use them to make our lives vastly better. The cost of not developing artificial intelligent machines and systems or forcing a slowdown in its development would result in slower progress in many important areas such as health care and our environment as well as in a wide array of goods and services.”

— — — — — — —

So let me briefly share my own views with you.

The recent advances in computer technology, that is frequently referred to as artificial intelligence has been nothing short of amazing during your and my lifetime. When I use Google Maps or talk to Siri on my I-phone I frequently use the term magic, even though I’m an engineer and understand some of the technology behind these amazing features. I believe that our rate of innovation is increasing and that this accelerated rate of change has many consequences that we can’t fully predict.

However there are two predictions that I feel reasonably comfortable in making. First, that the doomsday predictions regarding AI computer systems ruling the world and possibly threatening human existence is either science fiction or so far out in the distant future that it need not be a current concern. Yes, Stephen Hawking, Elon Musk and others are warning us but there are many other renowned scientists and engineers who see their concern as both an exceedingly low risk and in any case, a far far distant one in the future and one that we humans can prevent by establishing proper protective regulations and controls. In a recent interview Eric Schmidt, Google’s chairman said in response to the fears expressed by Musk and Hawkin: “Don’t you think humans would notice this happening? And don’t you think that humans would then go and simply turn those computers off ?”. I agree.

My second prediction is more of a concern and has a much closer time horizon.

Specifically, I am concerned about the possible impact of advanced computer systems on seriously dislocating of the labor force in both this country and around the world. AI systems may well result in a significant increase in unemployment both in the U.S. and in other parts of the world.

I realize that this concern is a complex subject all its own and one that has lots of political implications which I simply can’t address this morning, but it is an issue that governments around the world must evaluate and deal with or face the possibility of increasing social unrest.

There was an excellent article in the Sunday New York Times just two weeks ago under the title “The Real Threat of Artificial Intelligence”.

The author wasn’t concerned about a future doomsday scenario where AI computers took over the world, but he did express considerable concern over possible unemployment and economic disruptions. However even that concern is challenged by some who have studied this issue in detail. Deloitte, a U.S. consulting firm disputes the dire prediction of unemployment and points out that “While it is easy to point to jobs lost due to technology, it not as easy to identify the new jobs created by technology”.

— — — — — — —

Finally in closing, let me end on a positive note. It is clear that we stand on the brink of an amazing technological revolution that is fundamentally altering the way our children and their children will live, work, and how they will relate to one another. That transformation will be unlike anything humankind has previously experienced.

We don’t know just how it will unfold but the possibilities for humanity are staggering. Billions of people will be connected by a multitude of devices and have access to unlimited knowledge. And these possibilities will be further multiplied by continuing technology breakthroughs in , Robotics, the Internet of Things, Autonomous Vehicles, 3-D Printing, Nanotechnology, and even Quantum Computing. These are amazing tools that future generations will be able to use to solve the problems that they will encounter and to improve the quality of their life.

It’s an amazing world of the future that we will be passing on to our children and our grand children and I wish them well.

Jerry Seelig, 10 July, 2017