I appeared on the Dan Cofall Show Tuesday 10/7/14 to discuss my new book, The Artificial Intelligence Revolution (2014). Want to learn more about the merger of man and machine and the Singularity? There is also some disturbing news that machines may take up to 1/3 of jobs in the US. You can listen to my interview (5:00 PM CT segment) by clicking here. (Please give the page about 60-90 seconds to load)
Two days after the publication of my new book, The Artificial Intelligence Revolution (April 17, 2014), a short article authored by Stephen Hawking, Stuart Russell, Max Tegmark, and Frank Wilczek was published in the Huffington Post (April 19, 2014) under the title Transcending Complacency on Superintelligent Machines. Essentially the article warned, “Success in creating AI would be the biggest event in human history…Unfortunately, it might also be the last, unless we learn how to avoid the risks.” Shortly following the above publication, the Independent newspaper on May 1, 2014 ran an article entitled Stephen Hawking: Transcendence looks at the implications of artificial intelligence – but are we taking AI seriously enough?.
Recently, another notable artificial intelligence expert, Nick Bostrom, Professor in the Faculty of Philosophy at Oxford University. published his new book, Superintelligence (September 3, 2014), and offered a similar warning addressing the questions: ” What happens when machines surpass humans in general intelligence? Will artificial agents save or destroy us?”
It is unlikely that my book, which provides a similar warning and predates their warnings, was their impetus. I say unlikely because the time between the publication of my book, and it rise to number one on Amazon, is too close to their publications. Although, it is entirely possible that they may have read my book prior to going public with their warnings. The important point is that highly credible physicists and philosophers came to the same conclusion that formed the premise of my book: Artificial intelligence poses a potential threat to the long term existence of humankind.
Unfortunately, the artificial intelligence field is divided on the issue. Some believe that strong artificial intelligence, or what is also referred to as artificial general intelligence (i.e., intelligence equal to or exceeding human intelligence), will align itself with human goals and ethics. Ray Kurtzweil, a well know authority on artificial intelligence, acknowledges the potential threat exists, but suggests strong artificially intelligent entities will be grateful to humans for giving them existence. My view is this fraction is not looking at the facts. Humans have proven to be a dangerous species. In particular, we have set the world on a path of catastrophic climate change and engaged in wars, even world wars. We now have enough nuclear weapons to wipe out all life on Earth twice over. If you were a strong artificially intelligent entity, would you align with human goals and ethics? That is the crux of the issue. We pose a threat to strong artificially intelligent entities. Why wouldn’t they recognize this and seek to eliminate the threat?
Far fetched? Impossible? Consider this 2009 experiment performed by the Laboratory of Intelligent Systems in the Swiss Federal Institute of Technology in Lausanne. The experiment involved robots programmed to cooperate with one another in searching out a beneficial resource and avoiding a poisonous one. Surprisingly the robots learned to lie to one another in an attempt to hoard the beneficial resource (“Evolving Robots Learn to Lie to Each Other,” Popular Science, August 18, 2009). Does this experiment suggest the human emotion (or mind-set) of greed is a learned behavior? If intelligent machines can learn greed, what else can they learn? Wouldn’t self-preservation be even more important to an intelligent machine? Also, recognize that today’s robots would be eight times more intelligent that those in 2009, based on Moore’s law (i.e., computer technology doubles in capability every eighteen months).
The concern is real. We already have evidence (i.e., the Lausanne experiment) that suggest artificially intelligent entities will act in accordance with their own best interests and this does not have to be explicitly programmed. The evidence suggests, to my mind, that increased artificial intelligence gives rise to human-like mindsets like greed and self-preservation.
The time has come for the US to form an oversight committee to address the issue and suggest legislation. However, this issue is not just a US issue. It is a worldwide issue. For example, currently China has the most capable supercomputer (Tianhe-2). We must address this as a worldwide problem, similar to the way biological weapons and above ground nuclear testing was addressed. This means it must become one of the United Nation’s top issues.
There is a high urgency. Extrapolating today’s artificial intelligence technology, using Moore’s law, suggests that computes with artificial general intelligence will be built during the 2020 -2030 time-frame. Further extrapolation suggests computes that exceed the combined cognitive intelligence of all humans on Earth will be built in the 2040-2050 time-frame. The time to act is now, while humans are still the dominant species on the planet.
Can you computer become mentally ill? At first this may seem to be an odd question. However, I assure it is a potential issue. Let me explain further.
Most artificial intelligence researchers and futurist, including myself, predict that we will be able to purchase a personal computer that is equivalent to a human brain in about the 2025 time frame. Assuming for the moment that is true, what does it mean? In effect, it means that your new personal computer will be indistinguishable (mentally) from any of your human colleagues and friends. In the simplest terms, you will be able to carry on meaningful conversations with your computer. It will recognize you, and by your facial expressions and the tone of your voice it will be able to determine your mood. Impossible? No! In fact some researchers argue that machines should be able to interpret the emotional state of humans and adapt their behavior accordingly, giving appropriate responses for those emotions. For example if you are in a state of panic because your spouse is apparently having a heart attack, when you ask the machine to call for medical assistance, it should understand the urgency. In addition, it will be impossible for an intelligent machine to be truly equal to a human brain without the machine possessing human affects. For example how could an artificial human brain write a romance novel without understanding love, hate, and jealousy?
The entire science of “affective computing” (i.e., the science of programming computers to recognize, interpret, process, and simulate human affects) originated with Rosalind Picard’s 1995 paper on affective computing (“Affective Computing,” MIT Technical Report #321, abstract, 1995). In the last fourteen years, it has been moving forward. Have you noticed that computer generated voice interactions, such as ordering a new prescription from your pharmacy on the phone, is sounding more natural, more human-like? If you combine this information with the concept that to be equivalent to a human mine, the computer would also need to be self conscious.
You may argue if it is possible possible for a machine to be self-conscious. Obviously, since we do not completely understand how the human brain processes consciousness to become self-aware, it is difficult to definitively argue that a machine can become self-conscious or obtain what is termed “artificial consciousness” (AC). This is why AI experts differ on this subject. Some AI experts (proponents) argue it is possible to build a machine with AC that emulates the inter-operation of various parts of the brain called “neural correlates of consciousness” (NCC). NCC. Opponents argue that it is not possible because we do not fully understand the NCC. To my mind, they are both correct. It is not possible today to build a machine with a level of AC that emulates the self-consciousness of the human brain. However, I believe that in the future we will understand the human brain’s NCC inter-operation and build a machine that emulates it.
If in 2025 we indeed have computers equivalent to human minds, will they also be susceptible to mental illness? I think it is a possibility we should consider. We should consider it because the potential downside of a mentally ill computer may be enormous. For example, let’s assume we have a super intelligent computer managing the East Coast power grid. We replaced the human managers with a super intelligent computer. Now, assume the computer develops a psychotic disorder. Psychotic disorders involve distorted awareness and thinking. Two common symptoms of psychotic disorders are:
1. Hallucinations — the experience of images or sounds that are not real, such as hearing voices
2. Delusions — false beliefs that the ill person accepts as true, despite evidence to the contrary
What if our super intelligent computer managing the East Coast power grid believes (i.e., hallucinates) it has been given a command to destroy the grid and does so. This would cause immense human suffering and outrage. However, once the damage is done, what recourse do we have?
It is easy to see where I am going with this post. Today, there is no legislation that controls the level of intelligence we build into computers. There is not even legislation under discussion that would regulate the level of intelligence we build into computers. I wrote my latest book, The Artificial Intelligence Revolution (2014), as a warning regarding the potential threats strong artificially intelligent machines (SAMs) may pose to humankind. My point is a simple one. While we humans are still at the top of the food chain, we need to take appropriate action to assure our own continued safety and survival. We need regulations similar to those imposed on above ground nuclear weapon testing. It is in our best interest and potentially critical to our survival.
The short answer is “unlikely.” When the singularity occurs (i.e., when strong artificially intelligent machines exceed the combined intelligence of all humans on Earth), the SAMs (i.e., strong artificially intelligent machines) will use their intelligence to claim their place at the top of the food chain. The article “Is a Terminator-style robot apocalypse a possibility?” is one of many that have popped up in response to the my interview with the Business Insider (‘Machines, not humans will be dominant by 2045’, published July 6, 2014) and the publication of my book, The Artificial Intelligence Revolution (April 2014). If you would like a deeper understanding, I think you will find both articles worthy of your time.
Why should we be concerned about controlling the singularity when it occurs? Numerous papers cite reasons to fear the singularity. In the interest of brevity, here are the top three concerns frequently given.
There are numerous other scenarios, most of which boil down to SAMs claiming the top of the food chain, leaving humans worse off.
All of the above scenarios are alarming, but are they likely? There are two highly divergent views.
Whom should we believe?
It difficult to determine which of these experts accurately has predicted the postsingularity world. As most futurists would agree, however, predicting the postsingularity world is close to impossible, since humankind never has experienced a technology singularity with the potential impact of strong AI.
Martin believed we (humankind) may come out on top if we carefully handle the events leading to the singularity as well as the singularity itself. He believed companies such as Google (which employs Kurzweil), IBM, Microsoft, Apple, HP, and others are working to mitigate the potential threat the singularity poses and will find a way to prevail. He also expressed concerns, however, that the twenty-first century is a dangerous time for humanity; therefore he offered only a 50 percent probability that humanity will survive into the twenty-second century.
There you have it. Two of the top futurists, Kurzweil and Martin, predict what I interpret as opposing views of the postsingularity world. Whom should we believe? I leave that to your judgment.
Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte
Highly regarded AI researchers and futurists have provided answers that cover the extremes, and everything in between, regarding whether we can control the singularity. I will discuss some of these answers shortly, but let us start by reviewing what is meant by “singularity.” As first described by John von Neumann in 1955, the singularity represents a point in time when the intelligence of machines will greatly exceed that of humans. This simple understanding of the word does not seem to be particularly threatening. Therefore it is reasonable to ask why we should care about controlling the singularity.
The singularity poses a completely unknown situation. Currently we do not have any intelligent machines (those with strong AI) that are as intelligent as a human being let alone possess far-superior intelligence to that of humans. The singularity would represent a point in humankind’s history that never has occurred. In 1997 we experienced a small glimpse of what it might feel like, when IBM’s chess-playing computer Deep Blue became the first computer to beat world-class chess champion Garry Kasparov. Now imagine being surrounded by SAMs that are thousands of times more intelligent than you are, regardless of your expertise in any discipline. This may be analogous to humans’ intelligence relative to insects.
Your first instinct may be to argue that this is not a possibility. However, while futurists disagree on the exact timing when the singularity will occur, they almost unanimously agree it will occur. In fact the only thing they argue that could prevent it from occurring is an existential event (such as an event that leads to the extinction of humankind). I provide numerous examples of existential events in my book Unraveling the Universe’s Mysteries (2012). For clarity I will quote one here.
Nuclear war—For approximately the last forty years, humankind has had the capability to exterminate itself. Few doubt that an all-out nuclear war would be devastating to humankind, killing millions in the nuclear explosions. Millions more would die of radiation poisoning. Uncountable millions more would die in a nuclear winter, caused by the debris thrown into the atmosphere, which would block the sunlight from reaching the Earth’s surface. Estimates predict the nuclear winter could last as long as a millennium.
Essentially AI researchers and futurists believe that the singularity will occur, unless we as a civilization cease to exist. The obvious question is: “When will the singularity occur?” AI researchers and futurists are all over the map regarding this. Some predict it will occur within a decade; others predict a century or more. At the 2012 Singularity Summit, Stuart Armstrong, a University of Oxford James Martin research fellow, conducted a poll regarding artificial generalized intelligence (AGI) predictions (i.e., the timing of the singularity) and found a median value of 2040. Kurzweil predicts 2045. The main point is that almost all AI researchers and futurists agree the singularity will occur unless humans cease to exist.
Why should we be concerned about controlling the singularity when it occurs? There are numerous scenarios that address this question, most of which boil down to SAMs (i.e., strong artificially intelligent machines) claiming the top of the food chain, leaving humans worse off. We will discuss this further in part 2.
Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte
In our last post, we raised the question: “Will an intelligent machine ever be able to completely replicate a human mind?” Let’s now address it.
Experts disagree. Some experts—such as English mathematical physicist, recreational mathematician, and philosopher Roger Penrose—argue there is a limit as to what intelligent machines can do. Most experts, however, including Ray Kurzweil, argue that it will eventually be technologically feasible to copy the brain directly into an intelligent machine and that such a simulation will be identical to the original. The implication is that the intelligent machine will be a mind and be self-aware.
This begs one big question: “When will the intelligent machines become self-aware?”
A generally accepted definition is that a person is conscious if that person is aware of his or her surroundings. If you are self-aware, it means you are self-conscious. In other words you are aware of yourself as an individual or of your own being, actions, and thoughts. To understand this concept, let us start by exploring how the human brain processes consciousness. To the best of our current understanding, no one part of the brain is responsible for consciousness. In fact neuroscience (the scientific study of the nervous system) hypothesizes that consciousness is the result of the interoperation of various parts of the brain called “neural correlates of consciousness” (NCC). This idea suggests that at this time we do not completely understand how the human brain processes consciousness or becomes self-aware.
Is it possible for a machine to be self-conscious? Obviously, since we do not completely understand how the human brain processes consciousness to become self-aware, it is difficult to definitively argue that a machine can become self-conscious or obtain what is termed “artificial consciousness” (AC). This is why AI experts differ on this subject. Some AI experts (proponents) argue it is possible to build a machine with AC that emulates the interoperation (i.e., it works like the human brain) of the NCC. Opponents argue that it is not possible because we do not fully understand the NCC. To my mind they are both correct. It is not possible today to build a machine with a level of AC that emulates the self-consciousness of the human brain. However, I believe that in the future we will understand the human brain’s NCC interoperation and build a machine that emulates it. Nevertheless this topic is hotly debated.
Opponents argue that many physical differences exist between natural, organic systems and artificially constructed (e.g., computer) systems that preclude AC. The most vocal critic who holds this view is American philosopher Ned Block (1942– ), who argues that a system with the same functional states as a human is not necessarily conscious.
The most vocal proponent who argues that AC is plausible is Australian philosopher David Chalmers (1966– ). In his unpublished 1993 manuscript “A Computational Foundation for the Study of Cognition,” Chalmers argues that it is possible for computers to perform the right kinds of computations that would result in a conscious mind. He reasons that computers perform computations that can capture other systems’ abstract causal organization. Mental properties are abstract causal organization. Therefore computers that run the right kind of computations will become conscious.
Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte
Affective computing is a relatively new science. It is the science of programming computers to recognize, interpret, process, and simulate human affects. The word “affects” refers to the experience or display of feelings or emotions.
While AI has achieved superhuman status in playing chess and quiz-show games, it does not have the emotional equivalence of a four-year-old child. For example a four-year-old may love to play with toys. The child laughs with delight as the toy performs some function, such as a toy cat meowing when it is squeezed. If you take the toy away from the child, the child may become sad and cry. Computers are unable to achieve any emotional response similar to that of a four-year-old child. Computers do not exhibit joy or sadness. Some researchers believe this is actually a good thing. The intelligent machine processes and acts on information without coloring it with emotions. When you go to an ATM, you will not have to argue with the ATM regarding whether you can afford to make a withdrawal, and a robotic assistant will not lose its temper if you do not thank it after it performs a service. Highly meaningful human interactions with intelligent machines, however, will require that machines simulate human affects, such as empathy. In fact some researchers argue that machines should be able to interpret the emotional state of humans and adapt their behavior accordingly, giving appropriate responses for those emotions. For example if you are in a state of panic because your spouse is apparently having a heart attack, when you ask the machine to call for medical assistance, it should understand the urgency. In addition it will be impossible for an intelligent machine to be truly equal to a human brain without the machine possessing human affects. For example how could an artificial human brain write a romance novel without understanding love, hate, and jealousy?
Progress concerning the development of computers with human affects has been slow. In fact this particular computer science originated with Rosalind Picard’s 1995 paper on affective computing (“Affective Computing,” MIT Technical Report #321, abstract, 1995). The single greatest problem involved in developing and programming computers to emulate the emotions of the human brain is that we do not fully understand how emotions are processed in the human brain. We are unable to pinpoint a specific area of the brain and scientifically argue that it is responsible for specific human emotions, which has raised questions. Are human emotions byproducts of human intelligence? Are they the result of distributed functions within the human brain? Are they learned, or are we born with them? There is no universal agreement regarding the answers to these questions. Nonetheless work on studying human affects and developing affective computing is continuing.
There are two major focuses in affective computing.
1. Detecting and recognizing emotional information: How do intelligent machines detect and recognize emotional information? It starts with sensors, which capture data regarding a subject’s physical state or behavior. The information gathered is processed using several affective computing technologies, including speech recognition, natural-language processing, and facial-expression detection. Using sophisticated algorithms, the intelligent machine predicts the subject’s affective state. For example the subject may be predicted to be angry or sad.
2. Developing or simulating emotion in machines: While researchers continue to develop intelligent machines with innate emotional capability, the technology is not to the level where this goal is achievable. Current technology, however, is capable of simulating emotions. For example when you provide information to a computer that is routing your telephone call, it may simulate gratitude and say, “Thank you.” This has proved useful in facilitating satisfying interactivity between humans and machines. The simulation of human emotions, especially in computer-synthesized speech, is improving continually. For example you may have noticed when ordering a prescription by phone that the synthesized computer voice sounds more human as each year passes.
All current technologies to detect, recognize, and simulate human emotions are based on human behavior and not on how the human mind works. The main reason for this approach is that we do not completely understand how the human mind works when it comes to human emotions. This carries an important implication. Current technology can detect, recognize, simulate, and act accordingly based on human behavior, but the machine does not feel any emotion. No matter how convincing the conversation or interaction, it is an act. The machine feels nothing. However, intelligent machines using simulated human affects have found numerous applications in the fields of e-learning, psychological health services, robotics, and digital pets.
It is only natural to ask, “Will an intelligent machine ever feel human affects?” This question raises a broader question: “Will an intelligent machine ever be able to completely replicate a human mind?” We will address this question in part 2.
Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte
How and under what conditions is it possible for an intelligent machine to learn? To address this question, let’s start with a definition of machine learning. The most widely accepted definition comes from Tom M. Mitchell, a American computer scientist and E. Fredkin University Professor at Carnegie Mellon University. Here is his formal definition: “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.” In simple terms machine learning requires a machine to learn similar to the way humans do, namely from experience, and continue to improve its performance as it gains more experience.
Machine learning is a branch of AI; it utilizes algorithms that improve automatically through experience. Machine learning also has been a focus of AI research since the field’s inception. There are numerous computer software programs, known as machine-learning algorithms, that use various computational techniques to predict outcomes of new, unseen experiences. The algorithms’ performance is a branch of theoretical computer science known as “computational learning theory.” What this means in simple terms is that an intelligent machine has in its memory data that relates to a finite set of experiences. The machine-learning algorithms (i.e., software) access this data for its similarity to a new experience and use a specific algorithm (or combination of algorithms) to guide the machine to predict an outcome of this new experience. Since the experience data in the machine’s memory is limited, the algorithms are unable to predict outcomes with certainty. Instead they associate a probability to a specific outcome and act in accordance with the highest probability. Optical character recognition is an example of machine learning. In this case the computer recognizes printed characters based on previous examples. As anyone who has ever used an optical character-recognition program knows, however, the programs are far from 100 percent accurate. In my experience the best case is a little more than 95 percent accurate when the text is clear and uses a common font.
There are eleven major machine-learning algorithms and numerous variations of these algorithms. To study and understand each would be a formidable task. Fortunately, though, machine-learning algorithms fall into three major classifications. By understanding these classifications,we can gain significant insight into the science of machine learning. Therefore let us review the three major classifications:
In essence machine learning incorporates four essential elements.
Machine learning is similar to human learning in many respects. The most difficult issue in machine learning is generalization or what is often referred to as abstraction. This is simply the ability to determine the features and structures of an object (i.e., data) relevant to solving the problem. Humans are excellent when it comes to abstracting the essence of an object. For example, regardless of the breed or type of dog, whether we see a small, large, multicolor, long-hair, short-hair, large-nose, or short-nose animal, we immediately recognize that the animal is a dog. Most four-year-old children immediately recognize dogs. However, most intelligent agents have a difficult time with generalization and require sophisticated computer programs to enable them to generalize.
Machine learning has come a long way since the 1972 introduction of Pong, the first game developed by Atari Inc. Today’s computer games are incredibly realistic, and the graphics are similar to watching a movie. Few of us can win a chess game on our computer or smartphone unless we set the difficulty level to low. In general machine learning appears to be accelerating, even faster than the field of AI as a whole. We may, however, see a bootstrap effect, in which machine learning results in highly intelligent agents that accelerate the development of artificial general intelligence, but there is more to the human mind than intelligence. One of the most important characteristics of our humanity is our ability to feel human emotions.
This raises an important question. When will computers be capable of feeling human emotions? A new science is emerging to address how to develop and program computers to be capable of simulating and eventually feeling human emotions. This new science is termed “affective computing.” We will discuss affective computing in a future post.
Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte
Louis Del Monte on the Tom Barnard Show 7/23 discussing his new book, The Artificial Intelligence Revolution. During the interview we discuss the future of AI and how it may impact humanity. You can listen to the complete interview at anytime via this link http://www.tombarnardpodcast.com/july-23rd-2014-louis-del-monte-483-2/