#425 – BENEFITS AND RISKS OF AI – CAPERS JONES

Artificial intelligence (AI) is one of the most rapidly expanding technologies in human history.  It is also a disruptive technology that can do harm as well as good.   On the beneficial side are new methods or robotic surgery and improved citations in medical and scientific papers.

On the harmful and dangerous side, if a person’s voice has been recorded on the Web, criminals using AI can imitate the voice and send a message to the person’s family such as “help I’m being held for ransom.”  

AI can also be used for warfare and a new generation of pilotless tanks, aircraft, drones, and submarines are under construction that won’t require human crews.

This paper shows a number of uses of artificial intelligence and also the impact of AI on employment including:  1) the kinds of jobs that might be replaced by AI,  2) the kinds of jobs that will probably continue to be done by humans, and  3) new kinds of jobs due to AI itself.

Source: Pixabay

INTRODUCTION

As early as 1950 the scientist Alan Turing developed the “Turing test” to help identify artificial intelligence.   His test stated that if a mechanical device could carry on a conversation with a human that the human did not recognize as coming from a machine then the machine was a thinking machine or artificial intelligence device.

Today authors, publishers, and literary agents are facing competition from artificial intelligence. Artificial intelligence (AI) has demonstrated the ability to create short stories within 2 seconds.   Novels can be created in about 30 seconds, and a 500-page technical book can be created in about a minute.

In the future AI will be able to create computer software itself as well as supporting documents and technical manuals.

A full report on the impact of AI would be a book of more than 500 pages.  This short report covers a subset of AI capabilities.

There are over 20 AI tools available in 2023 such as ChatGPT, Speechify, Fireflies, ImagesAI for creating art, and many more.  Probably another dozen will be released by 2025.

Artificial intelligence is one of the most important inventions in human history and is just as important as the printing press, the computer, and the automobile.

BENEFICIAL USES OF ARTIFICIAL INTELLIGENCE

There are already dozens of beneficial uses of artificial intelligence and soon there will be many more, as discussed here.   Following are some hypothetical possibilities for the United States:

  1. Manufacturing of complex devices such as autos may be more than 90% computerized.
  2. 3D printed homes may approach 50% of U.S. home construction.
  3. Self-driving automobiles may approach 75% of U.S. auto sales.
  4. Robotic surgery may be used in over 75% of common medical operations.
  5. Artificial intelligence can be used to convert novels into movies automatically.
  6. Three-dimensional interactive movies might be possible.
  7. Music synthesizers will be able to duplicate all instruments and even orchestras.
  8. Military and commercial aircraft will have sophisticated autopilots.
  9. Robotic warships and robotic aircraft without crews will be added to arsenals.
  10. Urban traffic systems will synchronize traffic lights to optimize travel in crowded cities.
  11. AI can be used for customer support in many industries.
  12. AI can be used to teach elementary, high school, and college courses.
  13. AI will aid human authors in creating non-fiction research papers and books.
  14. AI will aid human fiction authors but may even compete with human authors.
  15. AI will be used by police and Federal investigators to collect evidence.

All of these changes due to AI will raise social issues that don’t exist in 2023.  The basic issue is what kinds of work will people do when computers have taken over many complicated jobs in fields such as home construction and manufacturing?

HARMFUL AND CRIMINAL USES OF ARTIFICIAL INTELLIGENCE

  1. Telephone scams by imitating human voices based on tapping telephone calls
  2. E-mail scams that imitate friends and family
  3. Computer scams that seek banking and personal information
  4. Spy satellites and spy balloons
  5. Creating imitations of famous art or fashion designs and passing them as originals
  6. Imitating voices of friends or colleagues in order to steal or get personal information
  7. Password guessing
  8. Fake news reports
  9. Targeted phishing
  10. Fake websites that offer low-cost items

The most important hazard is the use of AI for criminal activities.  If a person is well known and has photos on the web and any recordings of his or her voice, criminals could construct a virtual replica of the person and use it for extortion or theft.   It would easily be possible to have a phony AI replica place a telephone call to a relative or friend and ask for money.

Next it is interesting to consider the industries where computers in the future will either cause decreases in working populations or increases in working populations.   Following are brief discussions of the impact of AI on selected industries and occupations:

ARTIFICIAL INTELLIGENCE IN ARCHITECTURE

AI can aid human architects by speeding up designs and generating lists of materials.  In the future AI might also be able to construct homes.  Homes and office buildings are already being constructed using 3D printers.

ARTIFICIAL INTELLIGENCE FOR AUTHORSHIP AND PUBLISHING

Today 2023 AI is already well embedded in the creation and production of books and magazines:

Already in 2023 artificial intelligence can accomplish these writing topics which until recently could only be done by humans:

  1. Write novels.
  2. Write short stories.
  3. Write history books.
  4. Write technical manuals.
  5. Translate books and e-books into multiple natural languages.
  6. Create audio books.

Within a few years some additional AI capabilities should be possible:

  1. Convert novels into movies.
  2. Convert movies into novels.
  3. Convert novels or movies into television shows.
  4. Convert movies into 3D interactive movies.
  5. Develop unique scripts for new movies or television shows.
  6. Develop interactive movies where viewers can participate.

The screen writers’ strike of 2023 has stopped many TV shows from airing.   Unfortunately for the screen writers the ChatGPT AI tool could be used to generate scripts.  Protection from AI competition is one of the reasons for the strike.

Artificial intelligence in the future may have as large an impact on authorship and publishing as the printing press and the invention of e-books.  AI will also have major impacts on movies, television, and music composition and performances.

ARTICIAL INTELLIGENCE FOR PUBLISHING BOOKS AND MAGAZINES

Since AI can be used to generate movies from novels, it would also be possible to create audio books at the same time.   In fact more than one audio book could be produced with one having a male voice and the other a female voice.   Audio books could be produced in multiple spoken languages such as English, Japanese, French, German, etc.

In the future artificial intelligence (AI) might change the fundamental concept of novels from being fixed and static into a subscription model for e-books and related materials that evolve year by year for as long as the author and publisher want the process to continue.

Magazine publishing is expensive because of the large number of color illustrations.  AI could either produce the illustrations or possibly send out drones with cameras to photograph appropriate scenes.

Artificial intelligence can also create paintings that duplicate masters such as Cezanne.  AI can also produce images with photographic realism.   This raises a question about whether an AI version of a copyrighted artistic work would violate the copyright.

Magazines would also require minimal human staffing:  the owner, a human financial officer, and computer personnel.

If AI is used in publishing books and magazines, the jobs of thousands of humans would seem to be at risk.

Since magazines are heavily populated with photographs and images, AI can probably lower the costs of image creation to a small fraction of today’s costs.

ARTIFICIAL INTELLIGENCE FOR DISTRIBUTION OF PUBLICATIONS

Not only could artificial intelligence write stories and books, but AI can also edit the books, create art for books, print the books, bind the books, and package them for shipment.   AI could also handle orders for books.  AI could also create unique art for every book and thereby avoid copyright problems.

In fact, if AI is involved in book publishing to its maximum technical limits, the only humans involved in publishing would be:  1) the owner of the publishing company; 2) the chief financial officer, who has fiduciary responsibility that cannot be assigned to a machine; 3) three software engineers since AI publishing can operate 24 hours a day and a human should be available to fix computer problems if they occur.   If the publisher wants to stay open weekends, then additional software personnel would be needed.

Sales personnel would need to be human but could be contractors and not full-time employees.   Customer support is already handled by AI and only for issues that are outside the scope of the AI engine are humans involved.

If the publisher owns the AI computer then the publisher will need software and maintenance personnel.  But it is also possible to lease computer time from large vendors that support multiple companies.

Imagine a publishing company the size of McGraw Hill that would be able to produce over 1,000 new titles per year with only 5 to 7 human employees.

AI could produce so many books so quickly that it might saturate the markets.  However, AI can also do market analysis and would probably slow down production of new books to keep sales at reasonable levels.

The work of human authors requires editing and proofreading, which are not necessary for AI books.  It may be that in the future AI can do editing and proofreading of manuscripts written by human authors but this would put thousands of editors and proof readers out of business.

A question that goes beyond current knowledge would be what would happen if there were dozens or even hundreds of AI publishing companies?

Artificial intelligence in the future may have as large an impact on authorship and publishing as the printing press and the invention of e-books.  AI will also have major impacts on movies, television, and music composition and performances.

If an AI tool is set to imitate well-known authors such as Rudyard Kipling or Ernest Hemingway then AI could churn out dozens of sequels using the same style.  It will probably be necessary in the future for publishers to offer warranties that books were written by humans instead of by computers.

ARTIFICIAL INTELLIGENCE USED IN TELEVISION AND MOVIE PRODUCTION

Fairly soon artificial intelligence can be used for live television and movie production.  Some of the AI uses include set designs and art, background music composition, and controlling drones for filming outdoor backgrounds in hazardous locations such as the Arctic and the Amazon basin.   AI may also be used for creating scripts and story ideas.

The novel would have to be in e-book format.   Once published paper books do not change.  In fact some novels published on paper have lasted for over 1000 years.   If an author wants to create a new edition the whole book must be republished.

E-books, on the other hand, are downloaded into devices that can be connected to publishing websites.   If an author wants to make small changes to an e-book then the new version can be downloaded to any reader who wants the new version and has an e-book reader such as Amazon Kindle.

AI movies would compete with the current TV and Film production companies such as Amazon, Apple, Netflix, MGM, and many others.

Movies using images of famous actors and actresses would be more salable than movies using generic faces.   The only images that would be available without copyright violations would be those made before 1978.   It might be possible to use images of actors and actresses who are no longer living and whose work is out of copyright.

One very likely customer for movies produced automatically from e-books would be movie production companies such as Netflix or Amazon.   The automatic version would be useful in thinking about scenery, actions, and dialogs for later production of a live movie with modern human actors and actresses.

Should automatic production of movies from novels be of interest to book publishers or movie production companies, it would no doubt take several years before automatic conversion from novels to movies could be accomplished.

At a longer range it would be possible to convert novels into 3D movies and even to allow viewers to speak to characters in the movie version of the novel and get answers.    Not only that but publishers might allow simultaneous conversations among multiple readers of the same 3D animated novel.  Eventually it may be possible to produce 3D images on flat screens rather than using binocular devices.

There will be risks of copyright violations if images of real people are used in a computer-generated movie.   The risk sequence is:

  • The movie is an animated cartoon with no images of live people: low risk.
  • The movie shows generic images of people created by artificial intelligence.
  • The movie uses the author, friends, and relatives who give permission.
  • The movie shows actors and actresses who died before 1978: copyrights expired.
  • The movie shows modern actors and actresses who are paid to use their images.

The most likely first users of AI movies would be movie and TV production companies who use the computerized versions to experiment with camera angles, backgrounds, and other technical issues for lower costs than going on site.   Production companies are at as much risk as human authors in possibly losing jobs if artificial intelligence is used for movies and TV shows.

ARTIFICIAL INTELLIGENCE IN EDUCATION

AI can generate lectures at any level from grade school through post-doctorate levels.   AI can also give talks either using audio only or using images of professors.   One interesting use of AI might be a series of lectures using the images of the actual people who made important contributions to science and learning such as Charles Darwin, Albert Einstein, Hedy Lamarr, Thomas Edison, Nikolai Tesla, or Copernicus.

ARTIFICAL INTELLEIGENCE USES IN AVIATION AND MILITARY ACTIVITIES

One of the most important uses of artificial intelligence will be in the defense sector.  It is now possible to build aircraft, tanks, and even small surface ships that use AI instead of human crews.  This could greatly reduce casualties in warfare and military actions.     Already military drones use AI, and there are also submarine drones.

ARTIFICIAL INTELLIGENCE IN MANUFACTURING

One of the earliest and most widespread uses of artificial intelligence has been in manufacturing.  Robotic and AI controlled manufacturing are now being used for automobiles, aircraft, home appliances, medical devices, and industrial equipment.

ARTIFICIAL INTELLIGENCE IN MEDICAL PRACTICE

One of the major new uses of artificial intelligence will be in medical practice.   Already in 2023 AI is being used for diagnostic studies and for research into new kinds of medicines and treatments.   Robotic surgery is also becoming common.   It is also possible to build prosthetic devices such as artificial limbs controlled by computers using AI.

ARTIFICIAL INTELLIGENCE FOR COMPOSING AND PERFORMING MUSIC

One of the oldest uses of AI has been music performance.   AI can also compose music.  This means that not only are authors at risk from AI competition but also musicians and composers.

In the future it might be possible to generate a composition such as a piano concerto in a few seconds, and then have a performance of the concerto with avatars of the musicians.    The view from a computer screen would look like an actual auditorium with musicians and a director but the entire production would be artificial.

AI could also perform famous music by human composers such as Beethoven, Brahms, Puccini, Prokofiev, and any other composer so long as their music was written so long ago that it is not copyrighted.

ARTIFICIAL INTELLIGENCE AND SOFTWARE DEVELOPMENT

Within a few years artificial intelligence may be able to develop large systems in the 10,000- function point size range in less than 3 weeks instead of more than 3 calendar years which is the average for 2023.

In order for AI to generate large applications quickly and accurately the AI tool will need to have access to a large library of reusable components.   Some of these are available today from government sources.

AI application creation will also be able to generate user’s guides and tutorial materials to teach users how to provide inputs to the application and use all of the application’s functions and features.

AI can also handle management tasks before software development begins including but not limited to:

  • Predicting the size of the application in both lines of code and function points.
  • Predicting the numbers of organizations that will want to use the application.
  • Predicting probable updates for 3 years after deployment.

Once deployed AI maintenance tools can receive and repair bug reports, create new features as needed, and provide customer support for queries that come in via telephone or email.

The most likely organizations to create AI software development engines will be major computer and software companies such as IBM and Microsoft, government agencies such as the Department of Defense, and major corporations that use software for business purposes such as Amazon, General Motors, or Bank of America

SUMMARY OF BENEFICIAL AND HARMFUL USES OF ARTIFICIAL INTELLIGENCE

Future Beneficial Uses for Artificial Intelligence

  1. Author assistance for novels and scientific books
  2. Urban traffic controls and rerouting drivers away from blockages
  3. Routing emergency vehicles to fastest routes
  4. Continuous monitoring of medical patients in hospitals
  5. Converting books into television shows and movies
  6. Military threat recognition and target acquisition
  7. Pilotless combat aircraft
  8. Crewless submarines and naval patrol boats
  9. Pilotless reconnaissance drones
  10. Torpedo guidance and target seeking
  11. Building protection systems using night cameras and motion sensing
  12. AI-generated lectures at high school and university levels

Dangerous and Criminal Uses of Artificial Intelligence

  1. Telephone scams by imitating human voices based on tapping telephone calls
  2. E-mail scams that imitate friends and family
  3. Computer scams that seek banking and personal information
  4. Spy satellites and spy balloons
  5. Creating imitations of famous art or fashion designs and passing them as originals
  6. Imitating voices of friends or colleagues in order to steal or get personal information
  7. Password guessing
  8. Fake news reports
  9. Targeted phishing
  10. Fake websites that offer low-cost items
  11. Using AI to write term papers for university courses
  12. Using small AI controlled drones for industrial espionage and spying

The most important hazard is the use of AI for criminal activities.  If a person is well known and has photos on the web and any recordings of his or her voice, criminals could construct a virtual replica of the person and use it for extortion or theft.   It would easily be possible to have a phony AI replica place a telephone call to a relative or friend and ask for money.

Hopefully major software research companies such as IBM or Microsoft will be able to develop methods or tools that can detect whether a phone call is coming from an actual human or an imitation created by AI.    Until then millions of people are at risk from potential scams created by criminals using AI.

THE IMPACT OF ARTIFICIAL INTELLIGENCE ON THE U.S. WORKFORCE

There are over 1000 different occupations in the United States.  The last section of this report considers small samples showing only 15 jobs in three categories:    1) Possible job losses due to AI; 2) Jobs that cannot be performed by artificial intelligence; 3) New jobs created as a direct result of artificial intelligence.

U.S. JOBS THREATENED BY ARTIFICIAL INTELLIGENCE

2023 U.S. Employment = 165 million; threatened by AI = 105 million

  1. Accountants
  2. Advertising workers
  3. Authors
  4. Assembly line workers
  5. Banking
  6. Clerical workers
  7. College professors
  8. Customer support personnel
  9. Editors of books and magazines
  10. Musicians who record music
  11. Secretaries
  12. Software engineers on business applications
  13. Taxi drivers
  14. TV and movie production personnel
  15. Technical writers

U.S. JOBS THAT MAY BE SAFE FROM ARTIFICIAL INTELLIGENCE

2023 U.S. Employment = 165 million; unthreatened = 55 million

  1. Attorneys
  2. Plumbers, mechanics, etc.
  3. Coast Guard personnel
  4. Construction workers
  5. Elected officials: city, county, State, and Federal
  6. Electricians
  7. Firemen
  8. Flight attendants
  9. Government agencies such as the CIA, FBI, Internal Revenue, etc.
  10. Medical and dental personnel
  11. Military officers and personnel
  12. Police
  13. Sports: football, baseball, basketball, players and coaches, etc.
  14. Union members whose contracts prohibit AI replacement.
  15. Waiters and waitresse

NEW KINDS OF JOBS CREATED FOR USE OF ARTIFICIAL INTELLIGENCE

2023 U.S. Employment = 260 million; new kinds of jobs = 5 million

  1. AI assisted cyber-security experts.
  2. AI assisted commercial airline pilots
  3. AI assisted crews on freighters and large ships
  4. AI assisted movie and television production personnel.
  5. AI assisted home security specialists.
  6. AI assisted bankers and financial planners.
  7. AI assisted military planning personnel.
  8. AI assisted medical and hospital staff.
  9. AI assisted Federal and State investigative bureaus.
  10. AI assisted traffic control personnel.
  11. AI assisted commercial artists.
  12. AI assisted musical composers.
  13. AI assisted home and office building construction crews.
  14. AI assisted State and local government agencies.
  15. AI assisted tax collection agencies at all levels.

SUMMARY AND CONCLUSIONS ABOUT ARTIFICIAL INTELLIGENCE (AI)

In 2023 it is premature to know whether the impact of artificial intelligence will be mainly beneficial or mainly harmful to humans, or possibly both.

Artificial intelligence is a two-edged sword that can provide great benefits if used wisely and cause great harm if used for illegal or unethical purposes.    It is interesting that some of the pioneers of artificial intelligence are now trying to limit the expansion of AI.

AI is beneficial if used for purposes such as robotic surgery or expanding the references for scientific research papers.

AI is harmful if it is used to carry out criminal acts, to falsely claim authorship, or to cheat on college entrance examinations or  term papers.

CAPERS JONES’ GOVERNMENT CLIENTS FOR RISK CONSULTING

The author was an international risk consultant for IBM and his own company Namcook Analytics.   He has consulted about risks with over 75 corporations such as Apple,  AT&T, Ford, General Motors, and Microsoft.  He also consulted about risks with the government agencies shown here:

  1. Atomic Energy Commission (AEC)
  2. Department of Defense (DoD)
  3. Homeland Security
  4. Internal Revenue Service (IRS)
  5. National Aeronautics and Space Administration (NASA)
  6. National Security Agency (NSA)
  7. S. Airforce
  8. S. Navy
  9. Government of Canada
  10. Government of Hong Kong
  11. Government of Malaysia
  12. Government of Singapore
  13. Government of South Korea
  14. Government of Quebec
  15. Government of Thailand
  16. State of California
  17. State of New York
  18. State of Oregon
  19. State of Pennsylvania
  20. State of Rhode Island
  21. State of South Carolina

RECENT REPORT AND ARTICLES ON ARTIFICIAL INTELLIGENCE

Five Ways Artificial Intelligence will Change the World       NBC News 2023

The Future of AI’s Impact on Society                                     MIT Technology Review 2023

How Will AI Impact the Future of Work                                Forbes 2023

Artificial Intelligence News                                                   Science Daily 2023

Artificial Intelligence (AI) Technology                                  The Guardian 2023

Artificial Intelligence                                                              BBC News 2023

BOOKS ABOUT ARTIFICIAL INTELLIGENCE

The McGraw Hill Illustrated Encyclopedia of Robotics and Artificial Intelligence, McGraw Hill 2022.

Fundamental of Artificial Intelligence: Problem Solving and Automated Reasoning;  Miroslav Kubert, McGraw Hill, 2023

The Essence of Artificial Intelligence; Alison Crowly, Prentice Hall; 2023

Philosophy & Artificial Intelligence;  Todd C. Moody; Prentice Hall 2023

Artificial Intelligence:  A Modern Approach; Stuart Russel and Peter Norvig; Pearson; 2022

READINGS ON HISTORICAL TOPICS THAT TRANSFOMED AMERICA

Starr, Paul; The Social Transformation of American Medicine; Basic Books; Perseus Group; 1982; ISBN 0-465-07834-2.  NOTE:  This book won a Pulitzer Prize in 1982 and is highly recommended as a guide for improving both professional education and professional status.  There is much of value for the software community.

Strassmann, Paul; Information Payoff; Information Economics Press, Stamford, Ct; 1985.

Strassmann, Paul; Governance of Information Management: The Concept of an Information Constitution; 2nd edition; (eBook); Information Economics Press, Stamford, Ct; 2004.

Strassmann, Paul; Information Productivity; Information Economics Press, Stamford, Ct; 1999.

Weinberg, Gerald M.; The Psychology of Computer Programming; Van Nostrand Reinhold, New York; 1971; ISBN 0-442-29264-3; 288 pages.

Weinberg, Gerald M; Becoming a Technical Leader; Dorset House; New York; 1986; ISBN 0-932633-02-1; 284 pages.

Yourdon, Ed; Death March – The Complete Software Developer’s Guide to Surviving “Mission Impossible” Projects; Prentice Hall PTR, Upper Saddle River, NJ; ISBN 0-13-748310-4; 1997; 218 pages.

PUBLISHED BOOKS ABOUT RISK BY CAPERS JONES

  1. Software Development Patterns and Anti-Patterns; Taylor Francis, 2022
  2. Software Assessments, Benchmarks, and Best Practices; Addison Wesley Longman, Boston, Ma; 2000
  3. The Year 2000 Software Problem, Addison Wesley Longman, Boston, MA; 1998.
  4. Software Quality – Analysis and Guidelines for Success (International Thomson Computer Press; Boston, MA; 1997).
  5. Patterns of Software System Failure and Success; International Thomson Computer Press: Boston, MA; 1995).
  6. Assessment and Control of Software Risks; Prentice Hall, Englewood Cliffs, NJ; 1994.
  7. Applied Software Measurement; McGraw Hill, 1991
  8. Programming Productivity: Steps Toward a Science; Prentice Hall, 1993
  9. Software Quality Today; IBM Corporation; 1978
  10. Software Engineering Best Practices; IBM Corporation 1978

 JOURNAL ARTICLES ABOUT RISKS BY CAPERS JONES

Capers Jones has published more than 200 journal articles between 1978 and 2023 in magazines such as Scientific American, IBM Systems Journal, Datamation, Crosstalk, IEEE Transactions on Software Engineering, Cutter Software Journal, and others. This list primarily shows a sample of articles about risks.

  1. “High Efficiency Defect Removal Efficiency”; IEEE Software; August 2019
  2. “Challenges of Software Project Management”; IEEE Computer; June 2017”
  3. “Corporate Software Risk Reduction”; ITT Journal; August 2016
  4. “Defenses Against Software Litigation”; IEEE Computer; March 2015
  5. “Quality Control for Embedded Software”; IEEE Computer, May 2009.
  6. “Preventing Software Failure: Problems Noted in Breach of Contract Lawsuits”; U.S. Air Force software journal Crosstalk, June 2008.
  7. “Software Defect Potentials”; Crosstalk, Air Force Technology Support Center; December 2007
  8. “The Evolution of Defense Software”; Crosstalk; Air Force Technology Support Center; November 2004
  9. “Conflict and Litigation between Software Clients and Developers”; IEEE Computer; April 2001.
  10. “Analysis of Damages due to the Y2K Problem”; IEEE Software; December 2000.

Copyright © 2023 by Capers Jones.  All rights reserved.

BIO:

Capers Jones, Vice President and CTO, Namcook Analytics LLC
Email: Capers.Jones3@Gmail.com
Web:   www.Namcook.com

Leave a Reply

Your email address will not be published. Required fields are marked *