|
|
||||
|
| ||||
|
* * * * * * * * * |
* * Please note: This is the initial essay on Artificial Intelligence from 2018. It is where the discussion left off on the topic and has been posted here so readers won't have to search it out from last year. The discussion of the subject will continue a little later this summer. * *
The first essay of Summer 2019 will be available a little later this summer. The topic of Artificial Intelligence was started in 2018, and the discussion will continue this year. The essay archives currently are being revised and updated for 2019 - it's quite a time-consuming task, so thank you for your patience. The revised edition probably will not be available until later in the summer. The essay archives contain
all essays which have appeared on the site in the past.To visit the Essay Archives 2017, the most current edition of the archives, go
to www.dorothyswebsite.org/2017essay_archives.html. And, as always, thank you for your patience.
* * * * * * NEWS! NEWS! NEWS! - For drone fans, you can find the 2017 "A Bird's Eye View - The New Drone Universe" in the Essay Archives!
BIRD BRAINS: AN INTRODUCTION TO ARTIFICIAL INTELLIGENCE Artificial Intelligence, or AI, and terms associated with AI, have become increasingly prevalent not only in discussions of technology but also in discussions of everything from politics and power to business and employment (and even God and religion). Topics related to AI are sometimes presented in extreme polarities - that AI will either produce Terminator-like robots who will eventually destroy humanity or it will be the cure-all for all the world's most vexing problems. As with most discussions of complex topics, nothing is quite that simple. Though most people may not realize it, elements of AI already are present in their lives in everything from "personal assistants" like Siri or Alexa to talk of self-driving cars. What most experts agree upon, however, is that AI will come to permeate virtually every aspect of our lives in the coming decades. But what exactly is AI? This summer's series will provide an introduction to the topic and examples of how it is being used in a two-part series. The challenge is how to make discussion of the topic as relevant to someone who may have no knowledge of the term or subject as to someone who may have some expertise in the field. As with past essays, the story will be interactive in the sense that visitors may simply read the text or instead follow links, watch videos or look over photos and diagrams to gain a more in-depth understanding of the topics presented. Over the years a greater number of videos have become embedded in the essays as a popular means of being able to expand examples of what is being discussed in the text. When videos are included, every effort is made to present as thourough an example as possible of what is being discussed in a video which is under 3 - 5 minutes. While there are excellent explanations of the subject and related topics in videos 15 - 20 minutes (or longer), it also is understood that the casual viewer of this series is not likely to have the time to sit for hours on end to gain a better understanding of the topic at hand. Footnotes and a bibliography will be included at the end of the essay, and a link to a basic glossary of a few AI-related terms also is included here. (CLICK HERE to reach the glossary.) Comments, corrections or other input from visitors to the site, especially those with expertise in the field, are always welcome. If you would like to make a comment on this article, please send an e-mail to moremusic@dorothyswebsite.org. This article should remain here until about mid-September when the second essay of the summer, plus the Essay Archives revised for 2018, should be available. Depending on who you are talking or listening to, the term Artificial Intelligence (AI) conjures up visions of everything from autonomous robots taking over the human race to super-intelligent computer systems able to solve all of humankind's most vexing problems. The term - and others associated with it - are increasingly entering into public conversation. However, this conversation is not only taking place in the worlds of science and technology, but also increasingly in the realms of politics, government, religion, business, art and even pop culture. Just what is being said about AI? The following are a few examples (with many statements expanded upon in videos) from some people with whom most readers are familiar. In a recent speeched beamed into Russian schools, Vladimir Putin said that "AI . . . is the future, not only for Russia but for all humankind . . . Whoever becomes the leader in this sphere will become the ruler of the world." (n1) In addition, it was written in a recent Fortune article
that ". . . Unappreciated is China's growing strength in AI. . . The Chinese government recently announced its intention to catch up to the U.S. by 2020, surpass it by 2025, and dominate the industries of AI by 2030 with the help of its homegrown tech giants Baidu, Alibaba and Tencent . . ." (n2) And Facebook CEO Mark Zuckerberg, generally considered to be a proponent of AI, also cited
AI as a tool for eliminating hate speech and terrorist propaganda in his recent testimony before the U.S. Senate (included in the video below).
Video: Mark Zuckerberg on Artificial Intelligence During Senate Testimony. Others familiar with the technology, however, have had more cautionary predictions. In a BBC interview, the late Stephen Hawking said "I think the development of full artificial intelligence could spell the end of the human race." (n3) (To see the full video, go to www.youtube.com/watch?v=fFLVyWBDTfo. The subject of AI comes in
about minute 4 of the 5 minute video.) In a 2016 Fox Business interview, Bill Gates said that "I do think we need to worry about AI." (n4) (To see the full video, go to www.youtube.com/watch?v=EmfrMKLwr3k. The subject of AI comes in
about minute 2 of the 3 1/2 minute video.) And just last year, in a discussion at the National Governors Association meeting, Tesla and Space X CEO Elon Musk called AI "a fundamental existential risk for human civilization." (See the full video below.) [A note on the term "existential risk:" Elon Musk has donated millions of dollars to the Future of Life Institute
(www.futureoflife.org.) The organization lists is mission as "to catalyze and support research and initiatives for safeguarding life and developing optimistic visions for the future, including positive ways for humanity to steer its own course considering new technologies and challenges." The group currently is focusing on keeping artificial intelligence
beneficial. In the segment of the organization's website devoted to the term 'existential risk' (www.futureoflife.org/background/existential-risk), the term is defined as "any risk that has the potential to eliminate all of humanity or, at the very least, kill large swaths of the global population, leaving the survivors without sufficient means to
rebuild society to current standards of living."]
Video: Elon Musk Issues Yet Another Warning Against Runaway Artificial Intelligence - CNBC. This conversation about AI can be found extending beyond the bounds of business and government into current magazines, literature and even pop culture and video games. In 2017, Time magazine published a special edition on
AI titled "Artificial Intelligence: The Future of Humankind" in which it was said that "Artificial Intelligence . . . is poised to dramatically change everything from how astronomers explore the edges of our universe to whether your
stereo understands you were asking for more John Lennon, not John Legend." (n5) Dan Brown, author of many best-selling books including The Da Vinci Code, gives AI a major role in his new book "Origin." In a 2017 interview on CBS This Morning, Brown said that,
"AI is something that fascinates me deeply, mainly because scientists can't agree on whether it is going to save us or kill us . . . A lot of scientists feel we're on the verge of a new Renaissance, that AI will solve the big problems of humanity - scarcity, overpopulation, pollution. And yet, a lot of scientists feel that it is so powerful it will
destroy us. And as evidence they remind us that our species has never created technology that we have not weaponized, and it would be naive to think that AI will be any different." (n6) But Brown also made what author Douglas Busvine called a more "provocative remark" about AI at last year's Frankfurt Book Fair. The remark was that
humanity "may with the help of artificial intelligence develop a new form of collective consciousness that fulfills the role of religion." (n7) For science fiction/Star Trek fans, that notion of "collective consciousness" might conjure up images of arch-nemesis/super-villians The Borg, cybernetic organisms who forcibly transformed or "assimilated"
individuals into their collective. The Borg Wikipedia entry notes that "The Borg have become a symbol in popular culture for any juggernaut against which 'resistance is futile' [Borg mantra]." (n8) Is AI that juggernaut? These types of comments have not escaped the attention of those actually involved in AI research and development of products involving AI technology, some of whom have expressed concerns that "the way AI is being portrayed in public
discourse may ultimately impede AI research." (n9) All of the comments above do little or nothing to explain what AI is or what it does. The remaining portion of this first summer essay will serve as a basic introduction to AI, followed later by an article explaining current and/or projected future uses of the technology. But
as yet another example of how pervasive discussion of AI or terminology associated with AI has become, an anime/video game trailer for Fate/Grand Order - First Order is included here. Fate/Grand Order is an online free-to-play role playing mobile game based on the Fate/Staynight visual novel and franchised by Type-Moon, developed by Delight Works and published by
Sony Music Entertainment Japan's Aniplex subsidiary. In 2017 it was the sixth highest-grossing mobile game, (n10), and as of July of this year, the game had grossed over $2 billion in revenue worldwide. (n11) When watching the trailer, look for the use of the term "Singularity," a subject which will begin the discussion in the
next section.
Video: Fate - Grand Order: First Order Trailer. The Singularity, A Short History of AI, and Game-Playing Machines A Few Notes on the Singularity In the Fate - Grand Order video, the "singularity" is cited as a possible cause for humanity's extinction. The term "singularity" is one which pops up frequently in discussions of AI, and it is generally considered to mean
a future time at which machine intelligence exceeds human intellectual capacity. The term is most often associated with inventor/futurist Ray Kurzweil, author of the 2005 book The Singularity is Near (the book also has the subtitle "When Humans Transcend Biology"). Part of the premise of the "singularity" is based Moore's Law, a 1970's computer term referring to the fact
that computer processing speeds/capabilities have doubled roughly every two years, though it also incorporates the accelerating pace of technological change in general. In his book, Kurzweil defines/describes the "singularity" by saying that it is: " . . . a future period during which the pace of technological change will be so rapid, its impact so deep, that human life will be irreversibly transformed. Although neither uptopian nor dystopian, this epoch will transform the
concepts that we rely upon to give meaning to our lives, from our business models to the cycle of human life, including death itself . . . The key idea underlying the impending Singularity is that the pace of change of our human-created technology is accelerating and its powers are expanding at an exponential pace . . . The list of ways computers can now exceed human capabilities is rapidly growing. Moreover, the once
narrow applications of computer intelligence are gradually broadening in one type of activity after another. For example, computers are diagnosing electrocardiograms and medical images, flying and landing airplanes, controlling the tactical decisions of automated weapons, making credit and financial decisions, and being given responsibility for many other tasks that used to require human
intelligence. The performance of these systems is increasingly based on integrating multiple types of artificial intelligence . . . The book will argue . . . that within several decades information-based technologies will encompass all human knowledge and proficiency, ultimately including pattern-recognition powers, problem-solving skills, and emotional and moral intelligence of the human brain itself . . . From
a strictly mathematical perspective, the growth rates will still be finite but so extreme that the changes they bring about will appear to rupture the fabric of human history. That, at least, will be the perspective of unenhanced biological humanity. The Singularity will represent the culmination of the merger of our biological thinking and existence with our technology, resulting in a world that is still human
but transcends our biological roots." (n12) Kurzweil's most recent book is a 2012 book titled "How to Create a Mind: The Secret of Human Thought Revealed," and since 2012 he has been employed by
Google. There is no question that some of his notions regarding the singularity have been influential, though they are by no means universally accepted. Even a brief survey of AI-related literature turns up contradictory views and opinions of the subject. Take the following two passages, for example. The first is from an article at Smithsonian.com in which the writer, after interviewing "more
than a dozen futurists, philosophers, scientists, cultural psychiatrists and tech innovators, . . . [devises five possible] scenarios for the year 2065, ten years after the singularity arrives." (n13). The fourth hypothetical scenario, which he has named "Resistance is Costly," is described as follows: " . . . Imagine that you've opted out of the AI revolution. Yes, there are full-AI zones in 2065, where people collect healthy UBIs [Universal Basic Income, a political/social term which has become increasingly attached to AI debates denoting a set income given to people by the government] and spend their time making movies, volunteering and traveling the
far corners of the earth. But, as dazzling as a superintelligent world seems, other communities will reject it. There will be Christian, Muslim and Orthodox Jewish districts in cities such as Lagos and Phoenix and Jerusalem, places where people live in a time before AI, where they drive their cars and allow for the occasional spurt of violence, things almost unknown in full AI zones. The residents of these districts retain
their faith and, they say, a richer sense of life's meaning. The second passage is from an academic journal called the Journal of Experimental and Theoretical Artificial Intelligence. In the 2014 article, the author discusses papers from a conference on the "Impacts and Risks of Artificial General
Intelligence" that took place at the University of Oxford, St. Anne's College, in December of 2012. The author adds a note on terminology in which he says the following: "It is characteristic that none of the authors in this volume uses the term 'singularity' to characterise future development of AI - in fact, we had only a single paper submission using this word in the title or subtitle. People prefer other more specific terms like 'intelligence explosion,"
'AGI,' 'superintelligence,' 'acceleration,' etc. It would appear 'singularity' is now pretty much discredited in academic circles - with the notable exception of Chalmers (footnoted reference) and the ensuing debate. The discussions about singularity are generally characterised by convicton and fervor, which support amateurism and vitriolic exchanges - even in academically respectable publications . . .
Singularity is associated by ideological techno-optimism, trans-humanism and predictions like those of Ray Kurzweil (esp. Kurweil, 2005; more recently Kurzweil, 2012) that ignore the deep difficulties and risks of AI, e.g. by equating intelligence and computing power. What was the 'Singularity Institute' is now called the 'Machine Intelligence Research Institute' (MIRI). 'Singularity' is on its way towards becoming, literally,
the trademark of a particular ideology, without academic credentials." (n15) A Short History if AI
The concept of AI was outlined in a 1950 paper by Alan Turing called "Computing Machinery and Intelligence." The term "artificial intelligence" was first coined in the mid-1950s and is credited to computer scientist John McCarthy. In 1956 he organized the Dartmouth Conference, or
the Dartmouth Summer Research Project on Artificial Intelligence. At the six-week conference held at Dartmouth College, a "small group of mathematicians and scientists met daily . . . [in] an extended brainstorming session, . . . propos[ing] to show that every aspect of learning and other feature of intelligence can in principle be so precisely described that a machine can be made to
simulate it." (n16) The conference is considered to be the start of AI as an academic discipline. Though initial advances in the field were made between the 1950s and 1970s, "the slow pace of advancement in the 1980s . . . became colloquially known as the 'AI winter.'" (n17) Even into the early 1990s the field went through ups and downs, as "some AI research problems proved more difficult than anticipated, and others proved insurmountable with the technologies of the time." (n18) A U.S. government report dates the current wave of progress in AI to about 2010, "driven by three
factors that built upon each other: the availability of big data from sources including e-commerce, businesses, social media, science and government; which provided raw material for dramatically improved machine learning approaches and algorithms; which in turn relied on the capabilities of more powerful computers." (n19) Also not stated directly here but included is the advent
of the internet and the ability to access virtually limitless data via the internet. Successes in
the field have come from within both industry and government. Within the U.S. government, for example, the Defense Advanced Research Projects Agency (DARPA, www.darpa.mil) organized Grand Challenge competitions in the 2000s which contributed to advances in self-driving cars, and DARPA's Cognitive Agent that Learns and Organizes (CALO) led to Apple Inc.'s Siri. But
perhaps more useful in beginning to think about the evolution of AI is and what it can do are three examples of milestones which have come through challenges of machines vs humans in games. These three games are ones with which most people are familiar: chess, TV's Jeopardy! and most recently, go. Game-Playing Machines Deep Blue - Chess Last year marked the 20-year anniversary of one of the first major AI milestones, the 1997 victory of IBM computer Deep Blue over then world chess champion Garry Kasparov in a chess tournament. The subject was covered at length on this site during the 2011 "Game-Changer"-themed summer in a essay titled "A Game-Changing
Milestone: The Garry Kasparov-Deep Blue Chess Match and Beyond." [That essay currently can be found in the Essay Archives at July2011in17.html. Some of the material in this section is drawn from that essay; however, many of the web links valid at that time are no longer available. Currently, many older links to IBM Deep Blue materials redirect to the page
www-03.ibm.com/ibm/history/ibm100/us/en/icons/deepblue.] Two pioneers in the field of computing, Alan Turing of the University of Manchester in England, and Claude Shannon of MIT, first proposed the idea of chess-playing
computers after World War II and in the 1950s. But of all games, why chess? In genetics, the common fruitfly, drosophilia melanogaster, is considered an excellent subject for the study of broader genetic
questions. Chess is sometimes referred to as the "drosophilia" of games, and it was considered valuable for lessons which could be applied to both the study of artificial intelligence and real-world applications. The characteristics of the game that make it well-suited for
machine play are that "it is a game that pits one mind against another. There is no luck as in backgammon, no guessing of hidden information as in blackjack, [and] only a single opponent . . . [It is also] one of the few intellectual games where a fairly accurate performance measure exists [since] all strong
chess players in the world have ratings. (n20) Many of the elements of chess such as strategy, calculation, planning, etc. are all components of human intelligence which in some manner needed to be accounted for for a computer to be able to "play" the game. The challenge of developing a computer that could beat a world chess champion grew, and along the way it became something akin to the holy grail of AI at the time. The
problem and challenge attracted some of the best minds in computer science, and work was conducted at some of the nation's top universities and research laboratories. Gradual progress was made over the years, with computer programs gradually being
able to play against human opponents at higher and higher levels. The company which finally produced a computer which would go on to beat a human world champion was IBM. In 1989, "IBM hired a team of talented researchers who had just completed their Ph.D. studies at Carnegie Mellon University to design a program that would defeat the human world
champion, then Garry Kasparov. When the team joined IBM, it brought along its program, Deep Thought, which was playing at a weak grandmaster level." (n21) The first match between Deep Blue and Kasparov took place in February
of 1996 in Philadelphia, and Kasparov won 4-2. A rematch was set for May of 1997, and the IBM scientists went to work improving Deep Blue. What began as "an effort to explore how to use parallel processing [many processors working together
simultaneously] to solve computing problems" ended up becoming "what is . . . widely considered to be the greatest chess-playing computer ever constructed." (n22) In the year between the 1996 Philadelphia match and the 1997 rematch, the IBM scientists improved both Deep
Blue's speed and its chess knowledge. In 1996 Deep Blue could search "an average of 100 million [chess] positions per second," (n23) but the 1997 Deep Blue doubled that speed and was able to search 200 million positions per second (compared to 2 - 3 for Kasparov). (n24)
The scientists also increased Deep Blue's chess "knowledge" by using chess grandmasters to help in programming the system. The 1997 match took place in midtown Manhattan, and Deep Blue was the ultimate victor. The quest for that first holy grail was finally over, but was there any deeper meaning or use for the
results? In a post-match review article, Yale University computer science professor David Gelernter said of
Deep Blue that "Deep Blue is a beautiful and amazing technological achievement. It is an intellectual milestone, and its chief meaning is this: that human beings are champion machine builders. All sorts of
activities that we thought could be done only by minds can in fact be done by machines, too, if the machine builders are smart enough . . . [But in the end] it plays the game for the same reason a calculator adds
or a toaster toasts: because it is a machine designed for that purpose." (n25) Current IBM materials note that some of the lessons learned and experience gained from Deep Blue eventually went on to be used in business/industry for data mining (gaining insight from information contained
in large databases), financial modeling in the financial services industry, and molecular dynamics for drug discovery in the pharmaceutical industry. (n26) Deep Blue was eventually retired to the Smithsonian Museum in Washington D.C., and in 1999 IBM developed
the Deep Computing Institute to build on what had been learned with Deep Blue. Watson - Jeopardy! Around the time of the 10th anniversary of Deep Blue's victory, IBM researchers set their sites on a new goal -- the challenge of
building a computer that could compete on par with humans on the TV game show Jeopardy! While Deep Blue was able to search hundreds of millions of positions to come up with a move in the standard allotted time of three minutes, IBM researchers determined that "to compete [in Jeopardy!] . . . at human champion
levels, a computer system would need to answer roughly 70 percent of the questions asked with greater than 80 percent precision in three seconds or less." (n27) To meet this challenge, a research team from IBM's Deep QA project, led by principal investigator David
Ferrucci, created "Watson" (named after the founder of IBM), an artificial intelligence computer system capable of answering questions posed in natural language. To consider the advance that was represented by the ability of a computer to answer questions posed in natural language, consider the
following quote: "A computer system that can directly and precisely answer natural language questions over an open and broad range of knowledge has
been envisioned by scientists and writers since the advent of computers themselves. Consider, for example, the "Computer" in Star Trek. Taken to its ultimate form, broad and accurate open-domain question
answering may well represent a crowning achievement for the field of Artificial Intelligence (AI). While current computers can store and deliver a wealth of digital content created by humans, they are unable to operate over it in human terms. The quest for
building a computer system that can do open-domain Question Answering [QA] is ultimately driven by a broader vision that sees computers operating more effectively in human terms rather than strictly computer terms. They should function in ways that
understand complex information requirements, as people would express them, for example, in natural language questions or interactive dialogs. Computers should deliver precise, meaningful responses and synthesize, integrate and rapidly reason over
the breadth of human knowledge as it is most rapidly and naturally produced -- in natural language text." (n28) Watson's debut on Jeopardy! took place on February 14, 15 and 16 of 2011 and will be discussed here, followed by information on the technology behind the computer system. IBM's Watson competed against two of Jeopardy's best-known
champions: all-time highest money winner Brad Rutter, and Ken Jennings, the former contestant with the longest winning streak, with Watson's avatar, or symbol, as the "face" for the computer's stage "persona." A picture of the three contestants in a practice match is included
below. As with its human competitors, Watson was not allowed any wireless, Ethernet, internet or other outside connections, and it received its clues electronically at exactly the same time as the human players saw them. Watson was also required to use the
same hand buzzer as the other contestants to buzz in and answer the questions, though Watson was wired with a mechanical device to push the button. The Jeopardy! challenge was televised, and Watson
ended up winning the contest and the one million dollar prize, which IBM donated to charity. IBM's Watson computer system competes against Jeopardy's two most successful and celebrated contestants, Ken Jennings (left) and Brad Rutter (right), in a practice match held during a press conference at IBM's Watson Research Center
in Yorktown Heights, New York, in January of 2011. The televised Jeopardy! tournament was held on February 14, 15 and 16 of 2011. Source: IBM Press Room Press Release/Photo. So what exactly is Watson, and how did the computer win? More detailed technical specifications are available in this site's original July 2011 essay, but according to IBM figures, Watson was basically over 100 times more powerful than Deep Blue. At the heart of Watson,
and IBM's DeepQA, is still what some have referred to as the "brute force" of massive parallelism, or the large number of servers clustered together and linked via a super high-speed
network, roughly the equivalent of over 2,800 separate computers each with a single processor. In addition to the power of the parallel system, several other engineering and software features
played significant roles in Watson's success. One of the most important of these was Apapche UIMA, "a standard framework for building applications that perform deep analysis on unstructured content, including natural language text, speech, images and video." (n29) On a single processor, Watson required two hours
to answer a question. UIMA-AS was the "principal structure for assembling, scaling-out [over all processors] and deploying its analytic content" (n30), allowing Watson to deliver answers in 1 - 6 seconds. Another significant difference between Deep Blue and Watson was in the task each was constructed to perform. Deep Blue searched for chess moves among possible moves in the confined space of a chess board. Watson, on the other
hand, was tasked "to do what was once considered the exlusive domain or human intelligence: rapidly answer and rationalize open-domain natural language questions confidently, quickly and accurately . . . [while operating] in the nearly limitless, ambiguous and highly contextual domain of human language
and knowledge. Watson [was also] tasked to understand and answer human questions and to know when it [did] and [didn't] know the answer - to assess its own knowledge and ability - something humans find relatively easy." (n31) This was accomplished through what IBM referred to as "pervasive confidence estimation." (n32) The Watson system contained "roughly 200 million pages of natural language content (equivalent to reading
one million books)." (n33) Once a clue had been revealed, Watson's search for an answer began. Based on the computer's interpretation of the question and the data located, Watson's top three responses to each clue were displayed on the answer panel along with the confidence level for each answer. If the top response exceeded the confidence
threshold, Watson buzzed in to answer. The accomplishments of Watson are not considered the attaining of a "holy grail" or an end-point in an area of AI, but rather another step forward in the field. Unlike Deep Blue, Watson is still very much in use at IBM. Current information on Watson can be found at www-03.ibm.com/ibm/history/ibm100/us/en/icons/watson and
www.ibm.com/watson, and IBM says of Watson that it is powered by the latest innovations in machine learning (an AI term which will be defined shortly). (n34) Though the video below is about seven minutes, it is recommended for viewing as a highly concise and well-explained example of what Watson is as an AI system and how it can be used.
Video: IBM Watson: How it Works. AlphaGo - Go What is go? For those unfamiliar with go, it is a game which originated in China as wei-ch'i (pronounced way chee). Some of the earliest mentions of the game have been found in Chinese writings dating from several centuries B.C. Over the years the game has become better known by its Japanese name, "go." The standard go board is ruled with 19 x 19 horizontal and vertical lines and
traditionally has been played with black and white stones. The primary object of the game is to surround vacant space on the board, with a secondary object of capturing opponent's stones. It is considered to be one of the world's most difficult games to master. Part of the difficulty in designing a computer to compete at the game
was the fact that the number of possible moves in go is a number greater than the number of atoms in the universe, a number which was only determined in early 2016. (n35) In January of 2016, a group of from Google's DeepMind published a paper in Nature titled "Mastering the Game of Go with Deep Neural Networks and Tree Search" outlining the methodology used to develop AlphaGo, a computer program which defeated the European go champion 5 - 0. It was an
achievement which prior to that time was "thought to be at least a decade away." (n36) Who or what is DeepMind (www.deepmind.com)? DeepMind was founded in London in 2010 and was aquired by Google in 2014. The company's website says it "is the world leader in artificial intelligence research and its application for positive impact," and continues by adding "We're on a scientific mission to push the boundaries of AI,
developing programs that can learn to solve any complex problem without needing to be taught how." (n37) The video below (again, 7 minutes, but in a very descriptive and informative format), show members of DeepMind discussing Alpha Go after defeating the European go champion, but just prior to its match with world champion go player Lee Sedol.
Video: The Computer That Mastered Go. In March of 2016 AlphaGo beat Lee Sedol, the best go player in the world, 4 - 1 in a match played in South Korea. But that wasn't the end of development for AlphaGo. About year and a half after that championship match, DeepMind published a new paper in Nature titled "Mastering the Game of Go Without Human Knowledge." The paper referred to three versions of AlphaGo, called AlphaGo Fan, the version used to defeat
the European go champion, the improved version AlphaGo Lee, which defeated the world champion, and the newest version, AlphaGo Zero, which moves the company closer to its stated goal of "developing programs that can learn to solve any complex problem without needing to be taught how." The article introduces AlphaGo Zero by saying: "A long-standing goal of artificial intelligence is an algorithm that learns, tabula rasa, superhuman proficiency in challenging domains. Recently AlphaGo became the first program to defeat a world champion in the game of go. The tree search in AlphaGo evaluated positions and selected moves using deep neural networks. These neural networks were trained by supervised
learning from human expert moves, and by reinforcement learning from self-play. Here we introduce an algorithm based solely on reinforcement learning, without human data, guidance or domain knowledge beyond game rules. AlphaGo becomes its own teacher: a neural network is trained to predict AlphaGo's own move selections and also the winner of AlphaGo's games. This neural network improves the strength of the tree search, resulting in higher quality move selection and stronger self-play in the next iteration. Starting
tabula rasa, our new program AlphaGo Zero achieved superhuman performance, winning 100-0 against the previously published, champion-defeating AlphaGo." (n38) In the span of about 20 years, game-playing AI systems have passed through and beyond the game of chess, with a computer requiring input from chess grandmasters to help program the rules of play, to go, now with an algorithm through which the computer can teach itself to play with no human intervention other than a
basic outline of the rules of the game. By looking through some of the quotes and comments and watching the videos above, the reader also has been introduced to some of the basic language and terminology of AI - machine learning, algorithms, general artificial intelligence, neural networks, etc. Before concluding this article, a few of those terms briefly will be defined along with a couple of other thoughts on the general subject of AI. Defining AI Part of the problem in talking about AI is that it is not just one thing. Some define it as "a computerized system that exhibits behavior that is commonly thought of as requiring intelligence. Others define AI as a system capable of rationally solving complex problems . . . [which can include] the following taxonomy: 1) systems that think like humans (e.g. cognitive architecture and
neural networks), 2) systems that act like humans (e.g. pass the Turing test via natural language processing: knowledge representation, automated reasoning, and learning), 3) systems that think rationally (e.g. logic solvers, inference, and optimization), and 4) systems that act rationally (e.g. intelligent software agents and embodied robots that achieve goals via perception, reasoning, learning, communicating, decision-making and acting)."
(n39) Probably the most important distinction in terminology is between weak, or narrow AI and strong, or general AI, sometimes referred to as Artificial General Intelligence, or AGI. Weak or narrow AI (which is what most of today's AI is) refers to some of the type of things discussed above - AI developed and used for a specific application like playing games, image recognition or self-driving cars. Though the underlying principles may have broader use, a system designed
for playing chess, for example, could not be applied interchangeably to something else like driving a car. Artificial General Intelligence, or strong or general AI is the type of future AI system that would in theory be able to exhibit "intelligent behavior at least as advanced as a person across a full range of cognitive tasks. Attempts to reach General AI by expanding Narrow AI solutions have made little headway over many decades of research . . . [and] the current consensus . . . is
that General AI will not be achieved for at least decades." (n40) This is the type of AI envisioned in scenarios such as the singularity. The word "algorithm" has been used extensively in the text, and it is a term which goes hand in hand with "machine learning". Machine learning "is one of the most important technical approaches to AI and the basis of many recent advances and commercial applications of AI. Modern machine learning is a
statistical process [think, perhaps of some of the images from the Watson video] that starts with a body of data and tries to derive a rule or procedure that explains the data or can predict future data. This approach - learning from data - contrasts with the older "expert system" approach to AI, in which programmers sit down with human domain experts to learn the rules and criteria used to make decisions, and translate those rules into
software code." (n41) The newer approach was demonstrated above in the development of the DeepMind computer playing go; the older approach in the Deep Blue computer playing chess. The final term which will be discussed here is deep learning or deep [neural] networks. "In recent years, some of the most impressive advancements in machine learning have been in the subfield of deep learning, also known as deep network learning. Deep learning uses structures loosely inspired by the
human brain, consisting of sets of units (or "neurons"). Each unit combines a set of input values to produce an output value, which in turn is passed on to other neurons downstream . . . Deep learning networks typically use many layers - sometimes more than 100 - and often use a large number of units at each layer, to enable the recognition fo extremely complex, precise patterns
in data." (n42) The two-minute video below give a visual depiction of the relationship among the terms artificial intelligence, machine learning and deep learning.
Video: Artificial Intelligence Explained: Unleashing the Next Wave. As perhaps the reader has been able to learn from the above examples, AI has the potential to address problems and create solutions in everything from business and government to medicine and transportation. Some of the current and proposed or envisioned future applications of AI will be discussed in the
August/September essay. The essay should be available about mid-September. Thank you for visiting the page and please come back again next month! * * * * * Thanks for visiting the "Essays" page this month. Come back again in July or August to check on the next essay in the series! * * FOOTNOTES
- The following are the footnotes indicated in the text in parentheses
with the letter "n" and a number. If you click the asterisk at the end of
the footnote, it will take you back to the paragraph where you left
off. n1 - Simonite, Tom. "For Superpowers, Artificial Intelligence Fuels New Global Arms Race," Wired, 09.08.17, viewed online May 2018 at
www.wired.com/story/for-superpowers-artificial-intelligence-fuels-new-global-arms-race. (*) n2 - Nguyen, Minh-Ha, "Commentary: Why Silicon Valley Can't Get Complacent About China," fortune.com, December 6, 2017. Viewed online June 2018 at
www.fortune.com/2017/12/06/ai-wechat-alibaba-baidu-tencent-flipkart-india. (*) n3 - BBC News, Interview with Stephen Hawking, December 2, 2014. Viewed online July 2018 at www.youtube.com/watch?v=fFLVyWBDTfo. (*) n4 - "Davos 2016: Bill Gates on the Record," Fox Business interview video, January 22, 2016. Viewed online July 2018 at www.youtube.com/watch?v=EmfrMKLwr3k. (*) n5 - Vella, Matt, "How AI is Transforming Our World," in Artificial Intelligence: The Future of Humankind, New York: Time Inc. Books, 2017, p. 5. (*) n6 - Interview/online text, http://www.cbsnews.com/news/dan-brown-on-god-and-artificial-intelligence-in-his-new-thriller-origin, viewed interview on CBS December 2017 and online text of
article by David Morgan in July 2018. (*) n7 - Busvine, Douglas, "Collective Consciousness to Replace God - author Dan Brown," Reuters, October 12, 2017. Viewed online July
2018 at http://www.yahoo.com/entertainment/collective-consciousness-replace-god-author-dan-brown-120033225.html. (*) n8 - Viewed online July 2018 at http://en.m.wikipedia.org/wiki/Borg_(Star_Trek). (*) n9 - Johnson, Deborah G. and Verdicchio, Mario, "Reframing AI Discourse," Mind and Machines, (2017) 27:575-590, p. 575. (*) n10 - Superdata Research Holdings, Inc., 2017 Year in Review, Digital Games and Interactive Media, superdataresearch.com, January 25, 2018, p. 8. (*) n11 - Clayton, Natalie, "Fate/Grand Order Rakes in Over $2 Billion in Revenue Worldwide," PocketGamer.biz, 7/16/2018. Viewed online July 2018 at
http://www.pocketgamer.biz/asia/news/68582/fategrand-order-rakes-in-over-2bn-in-revenue-worldwide. (*) n12 - Kurzweil, Ray, The Singularity is Near: When Humans Transcend Biology, New York: Penguin Group, 2005, pp. 7 - 9. (*) n13 - Talty, Stephan, "Hacking the Future," Smithsonian.com, April 2018, p. 36. (*) n15 - Muller, Vincent C., "Editorial: Risks of General Artificial Intelligence," Journal of Experimental and Theoretical Artificial Intelligence,, 2014, Vol. 26, No. 3, 297 - 301. p. 299. (*) n16 - Vella, Matt, "How AI is Transforming Our World," in Artificial Intelligence: The Future of Humankind, New York: Time Inc. Books, 2017, p. 5. (*) n17 - Bowser, Anne, Sloan, Michael, Michelucci, Pietro and Pauwels, Eleonore, Artificial Intelligence: A Policy-Oriented Introduction, Wilson Briefs: November 2017, Washington D.C.: Woodrow Wilson International Center for
Scholars (Wilson Center), November 2017, p. 2. (*) n18 - Executive Office of the President of the United States, National Science and Technology Council, Committee on Technology, Preparing for the Future of Artificial Intelligence, Washington DC: Office of Science
and Technology Policy, October 2016, p. 5. This Obama administration document is one of the first U.S. goverment policy documents concerning AI. It can be found online at https://obamawhitehouse.archives.gov/sites/default/files/whitehouse_files/microsites/ostp/NSTC/preparing_for_the_future_of_ai.pdf. (*) n20 - Newborn, Monty, Deep Blue: An Artificial Intelligence Milestone. New York: Springer-Verlag 2003, pp. 4-5. (*) n22 - Birsic, Dorothy, "A Game-Changing Milestone: The Garry Kasparov-Deep Blue Chess Match and Beyond," July 2011, available online at http://www.dorothyswebsite.org/July2011in17.html. The original quote citation was
"IBM Research, "Deep Blue: This 1.4 Ton 8-year-old Sure Plays a Mean Game of Chess," which at the time of writing was available online at
www.research.ibm.com/deepblue/meet/html/d.3.shtml. The document was viewed in July 2011, but the page link no longer is valid. (*) n23 - Cipra, Barry. "Will a Computer Checkmate a Chess Champion at Last?" Science, vol. 271, 2 February 1996, p. 599. (*) n24 - Birsic, Dorothy, citing IBM Research, "Deep Blue," originally viewed in July of 2011. See note at Footnote 22 above. (*) n25 - Gelernter, David, "How Hard is Chess?" Time, Vol. 149, No. 20, May 19, 1997, p. 72. (*) n26 - IBM, "Deep Blue: Transforming the World," viewed online July/August 2018 at www-03.ibm.com/ibm/history/ibm100/us/en/icons/deepblue/transform. (*) n27 - IBM Systems and Technology, Watson - A System Designed for Answers, IBM White Paper. Somers, New York: IBM, February 2011,
p. 3. In 2011 the white paper was available online (after completing short registration form) at www.14.software.ibm.com/webapp/iwm/web/signup.do?source=stg-600BE30W. (*) n28 - IBM DeepQA Project information. Viewed online July 2011 at
www.research.ibm.com/deepqa/deepqa.shtml. This link is not longer valid and in August 2017 redirects to the IBM page https://researcher.watson.ibm.com/researcher/view_group.php?id=2099, a page which does not contain the same information as in 2011.
See note at Footnote 22 above. (*) n29 - IBM DeepQA/Watson FAQs, viewed online July 2011 at www.research.ibm.com/deepqa/faq.shtml. See additional notes at Footnote 22 and Footnote 28 above. (*) n30 - IBM Sytems and Technology Watson White Paper, p. 4. See additional notes at Footnote 22 and Footnote 27 above. (*) n31 - IBM DeepQA/Watson FAQs, viewed online July 2011 at www.research.ibm.com/deepqa/faq.shtml. See additional notes at Footnote 22 and Footnote 28 above. (*) n32 - IBM Sytems and Technology Watson White Paper, p. 4. See additional notes at Footnote 22 and Footnote 27 above. (*) n33 - IBM DeepQA/Watson FAQs, viewed online July 2011 at www.research.ibm.com/deepqa/faq.shtml. See additional notes at Footnote 22 and Footnote 28 above. (*) n34 - IBM, "Watson: What is Watson," viewed online August 2018 at http://www.ibm.com/watson/about. (*) n35 - Moyer, Christoper, "How Google's AlphaGo Beat a Go World Champion," The Atlantic Magazine/TheAtlantic.com, March 28, 2016. Viewed online at
http://www.theatlantic.com/technology/archive/2016/03/the-invisible-opponent/475611. (*) n36 - Silver, David, Huang, Aja, Maddison, Chris J. et al (20 named authors), "Mastering the Game of Go with Deep Neural Networks and Tree Search," Nature 529, 484-489, 28 January 2016. (*) n37 - DeepMind, "About Us," viewed online August 2018 at http://www.deepmind.com/about. (*) n38 - Silver, David, Schrittwieser, Julian, Simonyan, Karen, et al (17 named authors), "Mastering the Game of Go Without Human Knowledge," Nature 550, 354-359 plus supporting materials, 19 October 2017, p. 354. (*) n39 - Executive Office of the President of the United States, National Science and Technology Council, Committee on Technology, Preparing for the Future of
Artificial Intelligence, pp. 6 - 7. (*) LINKS LIST - The following is a list of links external to the website found in the essay. 1. AI glossary - https://dzone.com/articles/ai-glossary (viewed online August 2018) 2. Video: Mark Zuckerberg on Artificial Intelligence During Senate Testimony - www.youtube.com/watch?v=KBGCnzmpiiA 3. Video: Stephen Hawking: "AI could spell the end of the human race." - www.youtube.com/watch?v=fFLVyWBDTfo 4. Video: Bill Gates: "I think we do need to worry about Artificial Intelligence." - www.youtube.com/watch?v=EmfrMKLwr3k 5. Future of Life Institute - www.futureoflife.org 6. Future of Life Institute "existential risk" definition - www.futureoflife.org/background/existential-risk 7. Video: Elon Musk Issues Yet Another Warning Against Runaway Artificial Intelligence - www.youtube.com/watch?v=KdTTeR4TyMc 8. Video: Fate - Grand Order: First Order Trailer - www.youtube.com/watch?v=OrHgZwqUSPg 9. Obama administration AI report: Preparing for the Future of Artificial Intelligence - https://obamawhitehouse.archives.gov/sites/default/files/whitehouse_files/microsites/ostp/NSTC/preparing_for_the_future_of_ai.pdf 10. U.S. Defense Advanced Research Projects Agency (DARPA) - www.darpa.mil 11. Current IBM Deep Blue page - www-03.ibm.com/ibm/history/ibm100/us/en/icons/deepblue 12. Current IBM Watson pages - www-03.ibm.com/ibm/history/ibm100/us/en/icons/watson and www.ibm.com/watson 13. Video: IBM Watson: How it Works." - www.youtube.com/watch?v=_Xcmh1LQB9I 14. Google's DeepMind - www.deepmind.com 15. Video: Artificial Intelligence Explained: Unleashing the Next Wave (Intel) - www.youtube.com/watch?v=vehXkgG3YcU BIBLIOGRAPHY - The Bibliography for the June/July essay is included below. Akman, Varol, "Introduction to the Special Issue on Philosophical Foundations of Artificial Intelligence," Journal of Experimental and Theoretical Artificial Intelligence 12 (2000) 247 - 250. BBC News, Interview with Stephen Hawking, YouTube posted video, December 2, 2014. Viewed online July 2018 at www.youtube.com/watch?v=fFLVyWBDTfo. Birsic, Dorothy, "A Game-Changing Milestone: The Garry Kasparov-Deep Blue Chess Match and Beyond," www.dorothyswebsite.org, July 2011. Available online at http://www.dorothyswebsite.org/July2011in17.html Bogost, Ian, "Why Zuckerberg and Musk are Fighting About the Robot Future," The Atlantic, July 27, 2017. Viewed online May 2018 at www.theatlantic.com/technology/archive/2017/07/musk-vs-zuck/535077. Bowser, Anne, Sloan, Michael, Michelucci, Pietro and Pauwels, Eleonore, Artificial Intelligence: A Policy-Oriented Introduction, Wilson Briefs: November 2017, Washington D.C.: Woodrow Wilson International Center for Scholars (Wilson Center), November 2017. Busvine, Douglas, "Collective Consciousness to Replace God - author Dan Brown," Reuters, October 12, 2017. Viewed online July 2018 at http://www.yahoo.com/entertainment/collective-consciousness-replace-god-author-dan-brown-120033225.html. Cipra, Barry, "Will a Computer Checkmate a Chess Champion at Last?" Science, vol. 271, 2 February 1996, p. 599. Clayton, Natalie, "Fate/Grand Order Rakes in Over $2 Billion in Revenue Worldwide," PocketGamer.biz, 7/16/2018. Viewed online July 2018 at http://www.pocketgamer.biz/asia/news/68582/fategrand-order-rakes-in-over-2bn-in-revenue-worldwide. DeepMind, "About Us," (company description), viewed online August 2018 at http://www.deepmind.com/about. Della Cava, Marco, "Elon Musk Says AI Could Doom Human Civilization. Zuckerberg Disagrees. Who's Right?" USA Today, January 2, 2018. Viewed online May 2018 at www.usatoday.com/story/tech/news/2018/01/02/artificial-intelligence-end-world-overblown-fears/985813001. Executive Office of the President of the United States, National Science and Technology Council, Committee on Technology, , Preparing for the Future of Artificial IntelligenceWashington DC: Office of Science and Technology Policy, October 2016. Also online at https://obamawhitehouse.archives.gov/sites/default/files/whitehouse_files/microsites/ostp/NSTC/preparing_for_the_future_of_ai.pdf. Fox Business, "Davos 2016: Bill Gates on the Record," YouTube-posted interview video, January 22, 2016. Viewed online July 2018 at www.youtube.com/watch?v=EmfrMKLwr3k. Gelernter, David, "How Hard is Chess?" Time, Vol. 149, No. 20, May 19, 1997, p. 72. IBM, "Deep Blue: Transforming the World," viewed online July/August 2018 at www-03.ibm.com/ibm/history/ibm100/us/en/icons/deepblue/transform. IBM, "Watson: What is Watson," viewed online August 2018 at http://www.ibm.com/watson/about. IBM Systems and Technology, Watson - A System Designed for Answers, IBM White Paper. Somers, New York: IBM, February 2011. In 2011 the white paper was available online (after completing short registration form) at www.14.software.ibm.com/webapp/iwm/web/signup.do?source=stg-600BE30W. Johnson, Deborah G. and Verdicchio, Mario, , "Reframing AI Discourse," Mind and Machines, (2017) 27:575-590. Kurzweil, Ray, The Singularity is Near: When Humans Transcend Biology, New York: Penguin Group, 2005. Morgan, David, CBS News, interview/online text, http://www.cbsnews.com/news/dan-brown-on-god-and-artificial-intelligence-in-his-new-thriller-origin, viewed interview on CBS December 2017 and online text of article in July 2018. Moyer, Christoper, "How Google's AlphaGo Beat a Go World Champion," The Atlantic Magazine/TheAtlantic.com, March 28, 2016. Viewed online at http://www.theatlantic.com/technology/archive/2016/03/the-invisible-opponent/475611. Muller, Vincent C., "Editorial: Risks of General Artificial Intelligence," Journal of Experimental and Theoretical Artificial Intelligence,, 2014, Vol. 26, No. 3, 297 - 301. Muller, Vincent C., "Introduction: Philosophy and Theory of Artificial Intelligence," Minds & Machines (2012) 22: 67 - 69. Newborn, Monty, Deep Blue: An Artificial Intelligence Milestone, New York: Springer-Verlag 2003. Nguyen, Minh-Ha, "Commentary: Why Silicon Valley Can't Get Complacent About China," Fortune.com, December 6, 2017. Viewed online June 2018 at www.fortune.com/2017/12/06/ai-wechat-alibaba-baidu-tencent-flipkart-india. Silver, David, Huang, Aja, Maddison, Chris J. et al (20 named authors), "Mastering the Game of Go with Deep Neural Networks and Tree Search," Nature 529, 484-489, 28 January 2016. Silver, David, Schrittwieser, Julian, Simonyan, Karen, et al (17 named authors), "Mastering the Game of Go Without Human Knowledge," Nature 550, 354-359 plus supporting materials, 19 October 2017. Simonite, Tom, "For Superpowers, Artificial Intelligence Fuels New Global Arms Race," Wired, 09.08.17, viewed online May 2018 at www.wired.com/story/for-superpowers-artificial-intelligence-fuels-new-global-arms-race. Superdata Research Holdings, Inc., 2017 Year in Review, Digital Games and Interactive Media, superdataresearch.com, January 25, 2018. Talty, Stephan, "Hacking the Future," Smithsonian.com, April 2018, pp. 34 - 43. Torrance, Steve, "Producing Mind," Journal of Experimental and Theoretical Artificial Intelligence, 12 (2000) 353 - 376. Vella, Matt, "How AI is Transforming Our World," in Artificial Intelligence: The Future of Humankind, New York: Time Inc. Books, 2017, pp. 5 - 7. Wikipedia, Borg (Star Trek) entry, viewed online July 2018 at http://en.m.wikipedia.org/wiki/Borg_(Star_Trek).
Follow www.dorothyswebsite.org on TWITTER! Home |
Essays | Poetry | Free Concerts | Links | 2019 Extras |
About the Site |
|||
|
www.dorothyswebsite.org © 2003 - 2019 Dorothy A. Birsic. All rights reserved. Comments? Questions? Send an e-mail to: information@dorothyswebsite.org | ||||
