The next big area within the information and communication technology field is Artificial Intelligence (AI). The industry is moving to automate networks, cloud-based systems (e.g., Salesforce), databases (e.g., Oracle), AWS machine learning (e.g., Amazon Lex), and creating infrastructure that has the ability to adapt in real-time to changes and learn what to anticipate in the future. It is an area of technology that is coming faster and penetrating more areas of business than any other in our history. AI will be used from the C-suite to the distribution warehouse floor.
Replete with case studies, this book provides a working knowledge of AI's current and future capabilities and the impact it will have on every business. It covers everything from healthcare to warehousing, banking, finance and education. It is essential reading for anyone involved in industry.
Introduction to Artificial Intelligence and Machine Learning
Frank M. Groom
Ball State University
Contents
References
As the promise of Artificial Intelligence (AI) becomes the hot topic of the information technology world, Silicon Valley companies are allocating sizable resources toward developing and trialing their various products. Meanwhile giants such as Google are buying up promising AI companies such as AlphaGo which was created by the London-based company, DeepMind. AlphaGo then proceeded to beat the Go world champion Ke Jie (Russell, 2017)â. Meanwhile Ford, Tesla, and others are in the late development process of producing an automobile that drives itself.
These newsworthy incidents have heightened the interest of technocrats while simultaneously alerting the doomsayers. Apparently both are correct. AI and its partner, Machine Learning, are holding out enormous potential in all areas of business, communication, health, transportation, warehousing and delivery, on-line wholesale and retail, manufacturing, assembly, military, and even the last bastion government services. The benefits of automated decision-making and precision handling of objects have been gradually altering many areas of business from Amazonâs Fulfillment Centers to Mercedes, BMW, and Teslaâs virtually completely automated car assembly factories (McKinsey, 2017, 2018).
Although many believe there are limits to where AI can penetrate business operations, it increasingly appears that no job is safe from automation assistance and in some cases complete replacement (Bomey, 2017). In the 1990s, American businesses began the move of assembly and manufacturing work to the countries of Southeast Asia where labor rates were no more than 1%â3% of American rates, and as Labor Unions lost power, jobs were moved to the US Southern states and then on to Mexico and Southeast Asia. However, labor rates in those new assembly and manufacturing countries quickly began their own rise. Now China seeks to be the dominant player in offering Artificial Intelligent products to the world and is beginning to install such software in their own factories.
Initially it was thought that only the routine daily repetitive activities of white-collar work could be automated. But now with the demonstrated decision-making capabilities of IBMâs Deep Blue and Watson, and Googleâs AlphaGo, there is no reason to expect decision-making jobs are the exception. Even the areas that seemed impenetrable, such as writing and journalism, are being invaded. As The New York Times has noted, AI programs are widely being employed to sift through millions of items for financial news postings daily and summarizing them and even subsequently writing standard articles for direct posting to various news outlets and newspapers.
So where will the new jobs arise. Obviously the initial expectation is that there will be a dramatic demand for program coders, particularly those with the interest, intelligence, and coding skill to address AI and the enhancement of its capabilities by Machine Learning. The training of Machine Learning through thousands of actual examples augmented by continued learning from the Machine Learning Moduleâs actual live experiences has appealed to many potential programming hopefuls. Particularly attracted are the many game players and those who have devoted themselves to writing, sharing, and publishing their own games or variations on existing games. And the plentiful number of coding workshops, coding camps, and self-help on-line tutorials have set the stage for additional numbers who have interest to enter the programming community. Some will join the high-tech development community; some will become entrepreneurs and develop their own AI products. Others will join the business community as purchasers, analysts, implements, testers, and systems managers of the AI product line (Berbazzani, 2018).
However, there will be some limitations. With the emergence of Microsoftâs DeepCoder which searches through stacks of programmed routines and assembles programs from these libraries of code, Matej Balog at the University of Cambridge, along with Microsoft Researchâs Alexander L. Gaunt, Marc Brockschmidt, Sebastian Nowozin, and Daniel Tarlow, has begun the inevitable process of building AI coding programs that are tasked with assembling and writing, and required code on demand without the assistance of human coders.
Others say that human communication is the great barrier, but increasingly the initial experiments with automated help desks and other automated assistance have shown where such services can be applied as an option to deliver fast assistance where the wait for an available person can be lengthy and when available can be slow to understand, and even slower to determine how to fix the offered problem (Chui et al., 2016).
Decision-making will be the last bastion of human workers, but when the Hedge Fund Managers in Stamford Connecticut. It around while the programmed trading of most traffic on the stock exchanges under code provided by the mathematicians, physicists, engineers who inhabit the work; of the âQuantsâ, if our finances are underpinned by AI coded stock, bond, and futures trading, our day to day management decision-making seems rather mundane.
So our book intends to first present you with some of the basic elements of AI and Machine Learning, and then to show how it is currently being employed in a number of major industries in the Unites States. Figure 1.1 indicates the broad area covered by those who currently discuss AI. We discuss the general topic of deterministic systems which store the knowledge of experts and are triggered into action based upon a component of input that can be translated into an expertâs action. These were initially developed in the late 1980s as expert systems with much hope for utilization, but soon faded in the 1990s. Since then, beginning in the late 1990s and early 2000s, AI emerged as a topic of much interest in companies such as Google, particularly to solve their enormous search problems, and IBM, who saw potential to employ such programs on their large computers. These projects quickly advanced toward systems that could be trained with large amounts of data whose identity was known. Neuroscience added additional knowledge of wiring of neurons in human and animal brains (up to 100 billion neurons in humans, each with up to a thousand branches of axons that interconnect these neurons). From this insight, neural networks became the component of Machine Learning which can be taught with millions of live cases, and then begin learning on its own as it experiences live data in the situation where they are deployed. (Deep Learning then incorporated all the advancements resulting from creating increasing numbers of layers of intelligent nodes that could learn from their experiences (Aggarwal, 2018). Figure 1.1 portrays some of the components that comprise our current experiments with AI and Machine Learning.
Our book is organized with some initial chapters on the components and operation of AI and Machine Learning. These chapters follow the structure of the two books that are treated as the âbiblesâ in these areas: Stuart Russell and Peter Norvigâs Artificial Intelligence: A Modern Approach, (Russell & Norvig, 2010) which is the standard textbook in virtually all universities offering AI courses, and Ian Goodfellow, Yoshua Bengio, and Aaron Courvilleâs Deep Learning (Goodfellow et al., 2017), which is quickly assuming a similar stature covering Machine Learning. Other sources include Charu Aggarwalâs Neural Networks and Deep Learning (Aggarwal, 2018), Eugene Charniakâs Introduction to Deep Learning (Charniak, 2018), and Jeff Heatonâs three-volume Artificial Intelligence for Humans (Heaton, 2015). Furthermore, numerous papers on these subjects are available over the Internet including Vishal Maini and Samer Sabriâs very lucid Machine Learning for Humans (Maini & Sabri, 2017).
The initial basics chapters are followed by a series of chapters that address how and where AI is being employed by specific companies and industries. We have attempted to cover a broad spectrum of American businesses in order to demonstrate the importance and rapid deployment of resources to utilize such technologies and systems in their daily operation.
References
Aggarwal, Charu, Neural Networks and Deep Learning, Springer, Cham, Switzerland, 2018.
Berbazzani, Sophia, 10 Jobs Artificial Intelligence Will Replace (and 10 Jobs That Are Safe), HubSpot, Nov 7, 2018.
Bomey, Nathan, Special Report: Automation Puts Jobs in Peril, USA Today, Feb 6, 2017.
Charniak, Eugene, Introduction to Deep Learning, MIT Press, Cambridge, MA, 2018.
Chui, Michael, James Manyuika, and Mehdi Miremadi, Where Machines Could Replace Humans - And Where They Canât, McKinsey Quarterly, July 2016, https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/where-machines-âcould-replace-humans-and-where-they-cant-yet.
Heaton, Jeff, Artificial Intelligence for Humans, Vol 3 Deep Learning and Neural Networks, Heaton Research, Inc., Chesterfield, MO, 2015.
Goodfellow, Ian, Yoshua Bengio, and Aaron Courville, Deep Learning, MIT Press, Cambridge, MA, 2016.
Maini, Vishal and Samer Sabri, Machine Learning for Humans, August 19, 2017. https://medium.com/machine-learning-for-humans/why-machine-learning-matters-â6164fafldfl2.
McKinsey, Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation, McKinsey Global Institute, Dec 2017, https://www.mckinsey.com/~/media/mckinsey/featuredâ%20âinsights/future%20of%20organizations/what%20the%20future%20of%20work%20will%20mean%20for%20jobs%20skills%20and%20wages/mgi-jobs-lost-jobs-gained-report-december-6-2017.ashx.
McKinsey, Notes from the AI Frontier - Modeling the Impact of AI on the World Economy, McKinsey Global Institute, Sep 2018.
Russell, Jon, Googleâs AlphGo AI Wins a Three-Match Series Against the Worldâs Best Go Player, TechCrunch, May 5, 2017. https://techcrunch.com/2017/05/24/alphago-beats-planets-best-human-go-player-ke-jie/.
Russell, Stuart and Peter Norvig, Artificial Intelligence: A Modern Approach, 3rd ed., Pearson, Hoboken, NJ, 2010.
Chapter 2
The Basic Elements of Artificial Intelligence
Frank M. Groom
Ball State University
Contents
Introduction
Representation
Representing a Problem
Representation of Information
Storing Knowledge
Representing Knowledge
Frames
Trees
Objects
Formulating Problems and Solutions
Storing Knowledge and Information
File Structures
Storage Hardware
Searching for Knowledge and Actions
Breadth-First Search
Depth-First Search
Informed Search
Sophisticated Search
Heuristic Search
Local Search Algorithms
Adversarial Search and Games
Problem-Solving, Decision-Making, and Taking Action
Reasoning
Agents Employing Reasoning and Making Decisions
Resolution, Reaching Reasoned Conclusions, and Taking Appropriate Action
References
Introduction
As businesses recognize the breakthrough potential of Artificial Intelligence (AI) in the past decade, they have discovered that the theoretical possibilities are actually achievable with current software and hardware. From a software perspective, we have seen the IBM Deep Blue overcome world chess champion Garry Kasparov in 1997 after Kasparov won the first match in 1996. However, in 2017 Googleâs AI program, AlphaGo, beat Ke Jie who is arguably the best player of the most complicated game ever devised. AlphaGo was developed by DeepMind, the AI arm of Googleâs parent, Alphabet. Most Silicon Valley companies are placing large financial and personnel resources into not only AI research, but actually focusing on software and products that can be quickly brought to market.
Our newspapers delight in describing the successes and occasional tragedies emerging from the introduction of self-driving and assisted driving cars. The number of sensors, controllers, and communication modules offers a test-bed for a large number of AI components which the automotive industry is willing to finance. These automated components which the automotive industry is experimenting with will provide information which all technology companies will use to determine the level of control that localized and remote algorithmic software can exercise.
As these live experiments and deep research are employed, the decade of 2019â2029 is emerging as the anticipated time frame within ...