Social Machines:
The Coming Collision of Artificial Intelligence,
Social Networking, and Humanity
Apress, 2016
Summary of Book and Some Other Material
by Dr. Charles Tappert
Chap 1 – Introduction:
Why This Book

Two interwoven technologies are propelling the
increased incursion of machines into our lives

Social networks let us interact with other humans


With breakthroughs in AI, we often interact with computers


40% of those getting married now meet via online dating services
Siri, Alexa, and others becoming increasingly useful and ubiquitous
Authors


Both worked on a related DARPA research program
See a clash between AI, social networking, and humanity
because they worked on research that is forcing this collision
Chap 1 – Introduction:
Why This Book

Why read this book?



Learn about AI tools and products influencing us
Brief history of AI
Social machines

Term social machines used two ways in this book



Increasing ability of AI to allow computers to play human roles
Humans and computers together creating amazing systems
Risks and challenges


Potential for good but also potential to disrupt society
For example, the frightening view that AI will change how
we live, our society, maybe even end our lives
Chap 1 – Introduction:
Why This Book
Subject: Trump is actually making America great again
I can't believe I'm saying this, but it looks like Trump is actually making America great again. See progress since the election:...
1. Unprecedented levels of ongoing civic engagement.
2. Millions of Americans now know who their state and federal representatives are without having to google.
3. Millions of Americans are exercising more. They're holding signs and marching every week.
4. Alec Baldwin is great again. Everyone's forgotten he's kind of a jerk.
5. The Postal Service is enjoying the influx cash due to stamps purchased by millions of people for letter and postcard campaigns.
6. Likewise, the pharmaceutical industry is enjoying record growth in sales of anti-depressants.
7. Millions of Americans now know how to call their elected officials and know exactly what to say to be effective.
8. Footage of town hall meetings is now entertaining.
9. Tens of millions of people are now correctly spelling words like emoluments, narcissist, fascist, misogynist, holocaust, cognitive dissonance.
10. Everyone knows more about the rise of Hitler than they did last year.
11. Everyone knows more about legislation, branches of power and how checks and balances work.
12. Marginalized groups are experiencing a surge in white allies.
13. White people in record numbers have just learned that racism is not dead. (See #6)
14. White people in record numbers also finally understand that Obamacare IS the Affordable Care Act.
15. Stephen Colbert's "Late Night" finally gained the elusive #1 spot in late night talk shows, and Seth Meyers is finding his footing as today's
Jon Stewart.
16. "Mike Pence" has donated millions of dollars to Planned Parenthood since Nov. 9th.
17. Melissa FREAKING McCarthy.
18. Travel ban protesters put $24 million into ACLU coffers in just 48 hours, enabling them to hire 200 more attorneys. Lawyers are now heroes.
19. As people seek veracity in their news sources, respected news outlets are happily reporting a substantial increase in subscriptions, a boon to
a struggling industry vital to our democracy.
20. Live streaming court cases and congressional sessions are now as popular as the Kardashians.
21. Massive cleanup of Facebook friend lists.
22. People are reading classic literature again. Sales of George Orwell's "1984" increased by 10,000% after the inauguration. (Yes, that is true.
10,000%. 9th grade Lit teachers all over the country are now rock stars.)
23. More than ever before, Americans are aware that education is important. Like, super important.
24. Now, more than any time in history, everyone believes that anyone can be President. Seriously, anyone.
Chap 2 – Who Will Be
Your Next Doctor?

Cognitive Computing Technology




Computer systems that imitate how people solve problems
For example, IBM’s Watson playing Jeopardy
Clinical decision support systems for patient diagnosis
Other AI-based Systems in Healthcare




Image analysis for X-ray, ultrasound, CT scan, MRI
Web-based healthcare applications
Personalized healthcare applications
Wearable health-related computing devices
Chap 3 – The Games We Play


Explores differences between machines and
humans
Computers good at games with 3 properties:
1.
2.
3.

Perfect information game – no hidden information
Zero-sum – what’s good for one player is bad for the other
Deterministic – no luck involved
These games include

Tic-tac-toe, Chess, Go
Chap 3 – The Games We Play

Tic-tac-toe decision tree – initial portion
Chap 3 – The Games We Play

Chess – IBM’s Deep Blue beat world champion
Gary Kasparov in 1997
Chap 3 – The Games We Play

Go – Google’s AI Wins Fifth And Final Game
Against Go Genius Lee Sedol in 2016
Chap 3 – The Games We Play

Non-deterministic games (involve chance)


Backgammon – best players are computers
Hidden information games


Poker - A Mystery AI Just Crushed the Best Human
Players at Poker, 2017
Bridge
Chap 4 – The Limits of Humans



Human cognitive machinery limits our ability to
concentrate on a single thought deeply through
many, many alternatives
Pattern recognition – the ability to solve a new
problem or recognize a new pattern by
matching it to some previously solved problem
or observed pattern
Expert system building – interview (knowledge
acquisition) an expert and collect stories on
solving a specific type of problem
Chap 4 – The Limits of Humans



Humans are limited by physiological,
psychological, emotional, and social conditions
However, since the beginning of time humans
have created tools to augment their capabilities
Humans are now using cognitive computing
technologies to augment their capabilities
Chap 5
What Computers Can’t Do – Yet

Ambiguity in Language


Natural language understanding is not the same as
speech recognition (pure transcription)
For example, English has many words with multiple
meanings, usually disambiguated via context


“Time flies like an arrow” – several possible meanings
Language translation difficulties

English “The spirit is willing but the flesh is weak” when
translated into Russian and then back to English came
back “The vodka is strong but the meat is rotten”
Chap 5
What Computers Can’t Do – Yet

Ambiguity in Language
Chap 5
What Computers Can’t Do – Yet

Ambiguity in Language
Chap 5
What Computers Can’t Do – Yet

Ambiguity in Language


Natural language understanding is not the same as
speech recognition (pure transcription)
For example, English has many words with multiple
meanings, usually disambiguated via context


“Time flies like an arrow” – several possible meanings
Language translation difficulties

English “The spirit is willing but the flesh is weak” when
translated into Russian and then back to English came
back “The vodka is strong but the meat is rotten”
Chap 5
What Computers Can’t Do – Yet

Understanding the World We Live In


Necessary for natural language understanding
For example, English has many words with multiple
meanings, usually disambiguated via context


“Time flies like an arrow” – several possible meanings
Language translation difficulties

English “The spirit is willing but the flesh is weak” when
translated into Russian and then back to English came
back “The vodka is strong but the meat is rotten”
Chap 5
What Computers Can’t Do – Yet

The Problem of Context – three AI approaches
Procedural systems
1.


A program implemented algorithm solves the problem
Chess – minimax decision tree with alpha-beta pruning
Declarative systems
2.

Set of rules that generate actions (logic engine)
Learning systems
3.


Machine learning
Deep learning