Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
Artificial Intelligence is Maturing
1. 4/16/2018
1
David Smith
Artificial intelligence is maturing, can AI, IoT, sensors, Robotics, and
quantum computing lead to the next breakthrough for corporations?
Artificial intelligence is maturing, can AI, IoT,
sensors, Robotics, and quantum computing lead to
the next breakthrough for corporations? There
have been many recent advances in mathematical
algorithms, IoT, highly sensitive/compact sensors,
big data, mobile communications, and robotry.
The English mathematician Alan Turing gave the
first lecture on it in 1947. Is now the time
companies to go all in on AI?
Are we in the AI 3.0?
David Smith
Artificial intelligence is maturing, can AI, IoT,
sensors, Robotics, and quantum computing lead
to the next breakthrough for corporations? Are we
in the AI 3.0?
Copyright 2018 All Rights reserved May not be distributed without permission David Smith
Copyright 2018 David Smith All Rights Reserved
2. 4/16/2018
2
As we are in the new millennium science and
technology are changing rapidly
“Old” sciences such as physics
are relatively well-understood
Computers are ubiquitous
Grand Challenges in Science
and Technology
Understanding the brain
reasoning, cognition, creativity
creating intelligent machines
is this possible?
What are the technical and
philosophical challenges?
Arguably AI poses the most
interesting challenges and
questions in computer science
today
Copyright 2018 David Smith All Rights Reserved
3. 4/16/2018
3
ARTIFICIAL INTELLIGENCE
“AI is the study of
techniques for
solving exponentially
hard problems in
polynomial time by
exploiting knowledge
about the problem
domain.“
Elaine Rich
Definition
Artificial intelligence (AI) is intelligence exhibited
by machines. In computer science, the field of AI
research defines itself as the study of "intelligent
agents”
Artificial intelligence is technology that appears to
emulate human performance typically by
learning, coming to its own conclusions,
appearing to understand complex content,
engaging in natural dialogs with people.
The capability of a functional unit to perform
functions that are generally associated with
human intelligence such as reasoning and
learning. (ISO/IEC 2382-28:1995)
Copyright 2018 David Smith All Rights Reserved
4. 4/16/2018
4
What is Intelligence?
Intelligence:
- “The capacity to learn and solve problems” (Webster
dictionary)
- In particular,
• the ability to solve novel problems
• the ability to act rationally
• the ability to act like humans
Artificial Intelligence
- Build and understand intelligent entities or agents
- Two main approaches: “engineering” versus “cognitive
modeling”
Ex Machina Featurette - New Consciousness
Copyright 2018 David Smith All Rights Reserved
5. 4/16/2018
5
● Robotics is a major field related to AI.
● Robots require intelligence to handle tasks such as
object manipulation and navigation along with sub-
problems of localization, motion planning and
mapping.
Copyright 2018 David Smith All Rights Reserved
6. 4/16/2018
6
Philosophers have been trying for over 2000 years
to understand and resolve two Big Questions of the
Universe: How does a human mind work, and Can
non-humans have minds? These questions
are still unanswered.
Intelligence is the ability to understand and learn
things.
Intelligence is the ability to think and understand
instead of doing things by instinct or automatically.
Intelligent Machines, or
What Machines Can Do
(Essential English Dictionary)
What’s involved in Intelligence?
Ability to interact with the real world
- to perceive, understand, and act
- e.g., speech recognition and understanding and synthesis
- e.g., image understanding
- e.g., ability to take actions, have an effect
Reasoning and Planning
- modeling the external world, given input
- solving new problems, planning, and making decisions
- ability to deal with unexpected problems, uncertainties
Learning and Adaptation
- we are continuously learning and adapting
- our internal models are always being “updated”
• e.g., a baby learning to categorize and recognize animals
Copyright 2018 David Smith All Rights Reserved
7. 4/16/2018
7
Computers versus humans
A computer can do some things better than a human
can
- Adding a thousand four-digit numbers
- Drawing complex, 3D images
- Store and retrieve massive amounts of data
However, there are things humans can do much
better.
Thinking Machines
A computer would have
difficulty identifying the
cat, or matching it to
another picture of a cat.
Copyright 2018 David Smith All Rights Reserved
8. 4/16/2018
8
Or
AI Purposes
"AI can have two purposes. One is to use the power of
computers to augment human thinking, just as we use
motors to augment human or horse power. Robotics and
expert systems are major branches of that. The other is
to use a computer's artificial intelligence to understand
how humans think. In a humanoid way. If you test your
programs not merely by what they can accomplish, but
how they accomplish it, they you're really doing cognitive
science; you're using AI to understand the human mind."
- Herb Simon
Copyright 2018 David Smith All Rights Reserved
9. 4/16/2018
9
“We cannot solve our problems
with the same thinking we
used when we created them.”
- Albert Einstein
Overview of Artificial Intelligence
Definitions – four
major combinations
- Based on thinking
or acting
- Based on activity
like humans or
performed in
rational way
Systems
that think
like
humans
Systems
that think
rationally
Systems
that act
like
humans
Systems
that act
rationally
Copyright 2018 David Smith All Rights Reserved
10. 4/16/2018
10
“The market for enterprise AI systems will increase from $202.5 million
in 2015 to $11.1 billion by 2024.”
- Tractica
Copyright 2018 David Smith All Rights Reserved
11. 4/16/2018
11
Near Future
• By 2018, 20 percent of business content will be authored by
machines.
• By 2020, autonomous software agents outside of human
control will participate in five percent of all economic
transactions.
• By 2019, more than 3 million workers globally will be
supervised by a "robo-boss.“
• By 2018, 45 percent of the fastest-growing companies will
have fewer employees than instances of smart machines.
• By year-end 2018, customer digital assistant will recognize
individuals by face and voice across channels and partners.
• By 2020, smart agents will facilitate 40 percent of mobile
interactions, and the post app era will begin to dominate.
Classical Digital Computer
Moore’s Law: # of transistors on chip doubles every 18 months—
microprocessor circuits will measure on atomic scale by 2020
Downscaling of circuit board layout/components is leading to
discrepancies.
Copper traces are actually crystallizing and shorting out!
Emergence of quantum phenomena such as electrons tunneling through
the barriers between wires.
Serial Processing – one operation at a time
64-bit classical computer operates speeds measured in gigaflops (billions
of floating-point operations per second).
Quantum Computer
Harnesses the power of atoms and molecules to perform memory and
processing tasks
Parallel Processing – millions of operations at a time
30-qubit quantum computer equals the processing power
of conventional computer that running at 10 teraflops
(trillions of floating-point operations per second).
The Need For Speed...
Copyright 2018 David Smith All Rights Reserved
13. 4/16/2018
13
“A quantum computer is to a regular computer, what a
laser is to a lightbulb.”
--Seth Lloyd, MIT
One analogy…
“The thing driving the hype is the
realization that quantum computing is
actually real. It is no longer a physicist’s
dream—it is an engineer’s nightmare.”
“Nature is quantum, goddamn it! So if we
want to simulate it, we need a quantum
computer.”
Satya Nadella, Microsoft CEO: “The world is
running out of computing capacity. Moore’s law
is kinda running out of steam … [we need
quantum computing to] create all of these rich
experiences we talk about, all of this artificial
intelligence.”
Copyright 2018 David Smith All Rights Reserved
14. 4/16/2018
14
Google Rattles the Tech World With a New
AI Chip
Neuromorphic Computing And OTHERS
“Spikey” from
Electronic Visions
group in Heidelberg
Qualcomm’s NPU’s
for robots.
SpiNNaker’s 1B
neuron machine
Stanford’s Neurogrid
Intel’s concept design...
IBM’s
TrueNorth
(Peter Nugent, LBNL)
Copyright 2018 David Smith All Rights Reserved
15. 4/16/2018
15
Second Key to Creating Artificial General Intelligence:
Making It Smart
Strategies:
1) Plagiarize the brain.
• Reverse engineer it
• Build chips to simulate it
• Capture its synapses
• “Whole brain emulation”
2) Try to make evolution do what it did before but for us this time.
• Use foresight – just pick what you know will win
• Select for intelligence
• Provide externally what evolution takes extra steps to do, i.e.,
provide outside energy/electricity
3) Make this whole thing the computer’s problem, not ours
• It would do research on AI and code the changes into itself
Now: 1 mm-long
flatworm brain of
302 Neurons
Although artificial intelligence as an independent field of
study is relatively new, it has some roots in the past. We can
say that it started 2,400 years ago when the Greek
philosopher Aristotle invented the concept of logical
reasoning. The effort to finalize the language of logic
continued with Leibniz and Newton. George Boole
developed Boolean algebra in the nineteenth century
that laid the foundation of computer circuits. However, the
main idea of a thinking machine came from Alan Turing, who
proposed the Turing test. The term “artificial intelligence”
was first coined by John McCarthy in 1956.
History of artificial intelligence
Copyright 2018 David Smith All Rights Reserved
16. 4/16/2018
16
A Brief History of AI
5th century BC
Aristotle invents syllogistic logic, the first formal deductive
reasoning system.
16th century AD
Rabbi Loew supposedly invents the Golem, an artificial
man made out of clay.
17th century
Descartes proposes animals are machines and founds a
scientific paradigm that will dominate for the coming
centuries.
Pascal creates the first mechanical calculator in 1642
18th century
Wolfgang von Kempelen “invents” fake chess-playing
machine, The Turk.
19th century
• George Boole creates a binary algebra to represent
“laws of thought”.
• Charles Babbage and Lady Lovelace develop
sophisticated programmable mechanical computers,
precursor to modern electronic computers.
First Half of 20th century
• Karel Kapek writes “Rossum’s Universal Robots”, coining
the English word “robot”.
• Warren McCulloch and Walter Pitts lay partial
groundwork for neural networks.
• Turing writes “Computing Machinery and Intelligence” –
proposal of Turing test.
Copyright 2018 David Smith All Rights Reserved
17. 4/16/2018
17
March 2016: Introduction of Sophia, a learning
& social android with humanlike facial
expression, eye contact, hearing and vocal
qualities.
March 2016: Google AlphaGo, using deep
neural network and tree search to decide
and learn quickly, defeats world Go champ
Lee Se-dol 4 to 1.
Meet HAL
2001: A Space Odyssey
- classic science fiction movie from 1969
HAL
- part of the story centers around an intelligent computer called HAL
- HAL is the “brains” of an intelligent spaceship
- in the movie, HAL can
• speak easily with the crew
• see and understand the emotions of the crew
• navigate the ship automatically
• diagnose on-board problems
• make life-and-death decisions
• display emotions
In 1969 this was science fiction: is it still science fiction?
Copyright 2018 David Smith All Rights Reserved
18. 4/16/2018
18
AI 1.0 (1960-1985):
AI applications addressed a single area. In this period, they were high value such as human language
translation and route optimization centered around the high cost of humans. Algorithms were
mechanistic. Heavy demand for IT resources made implementations expensive. Today, single area AI
applications, enabled by more sophisticated mathematics and high performance computing, is labelled
Artificial Narrow Intelligence (ANI).
AI 2.0 (1986 - 2010):
AI applications appeared to address a broad area. In this period, they were capable of doing the work
of an occupation of people such as picking crops, scanning social networks for consumer input, and
classifying images for quicker retrieval. Algorithms became more sophisticated and IT resources much
less expensive. However, the solutions approach mimic how humans thought and still fell short of the
abilities of experts. This class of AI Application is labelled Artificial General Intelligence (AGI).
AI 3.0 (2011 - Now):
AI applications are appearing that can solve problems better than the best human in an area of
interest. Examples of this class of AI application can win a the most complex strategic board games,
perform retrieval and analysis of knowledge to quickly answer questions, and stock market trading.
This generational shift has been driven by high value potential, accumulation of massive data of all
kinds, even faster computers the ability to analyze a single situation across a cluster of computers,
and the algorithms to exploit the new technological resources to analyze problem deeper to
incorporate behavioral/neuro/ social data to perform real time analysis and even learn. This class of AI
application is being called Artificial Superintelligence (ASI).
Artificial Intelligence Generations
Copyright 2018 David Smith All Rights Reserved
19. 4/16/2018
19
The vast majority of AI research practiced in
academia and industry today fits into the “Narrow
AI” category
Each “Narrow AI” program is (in the ideal case)
highly competent at carrying out certain complex
goals in certain environments
• Chess-playing, medical diagnosis, car-driving,
etc.
Narrow AI
“The ability to achieve complex goals in
complex environments using limited
computational resources”
• Autonomy
• Practical understanding of self and
others
• Understanding “what the problem
is” as opposed to just solving
problems posed explicitly by
programmers
Artificial General Intelligence (AGI)
Artificial General Intelligence (AGI)
Copyright 2018 David Smith All Rights Reserved
20. 4/16/2018
20
Artificial Intelligence Generation Comparison
Factor Generation AI 1.0 AI 2.0 AI 3.0
Period of Time 1960 to 1985 1986 to 2010 2011 to Now and
beyond
Type of AI App
Introduced
Artificial Narrow
Intelligence (ANI)
Artificial General
Intelligence (AGI)
Artificial
SuperIintelligence (ASI)
Value Proposition Human Efficiency Human Effectiveness Human Substitution
Human Ability
Acquired
Fast manipulation of
text and data
Incorporation of
knowledge,
Audio/visual recognition
Understanding,
Reasoning
ANI Roadmap Batch processing Complex data/math Real time
AGI Roadmap Longitudinal data,
Pattern recognition
Data warehouses,
Non-SQL data bases
ASI Roadmap Deep Neural Nets,
Big Data, Robotics
Different Types of Artificial
Intelligence
Modeling exactly how humans actually think
- cognitive models of human reasoning
Modeling exactly how humans actually act
- models of human behavior (what they do, not how they think)
Modeling how ideal agents “should think”
- models of “rational” thought (formal logic)
- note: humans are often not rational!
Modeling how ideal agents “should act”
- rational actions but not necessarily formal rational reasoning
- i.e., more of a black-box/engineering approach
Modern AI focuses on the last definition
- we will also focus on this “engineering” approach
- success is judged by how well the agent performs
-- modern methods are also inspired by cognitive & neuroscience (how
people think).
Copyright 2018 David Smith All Rights Reserved
21. 4/16/2018
21
A Human vs. Machine Comparison
Category Attribute Man Machine
Hardware Processing speed Max @ 200 cycles/sec Already 2 billion cycs/sec
Interconnect speed ~ 120 meters/second Speed of light
Size/Storage Size of skull; any
bigger we’d think more
slowly
Greatly expandable in short
term/working/long term
memories; has error
detect/self-correct bits
Reliability/durability Get easily fatigued; will
deteriorate over time
Transistors more accurate
that neurons; can be
repaired or replaced; can
run non-stop 24/7
Software Programmability Human brain is not
“updatable”
Can be optimized to suit its
role; improvable; fixable
“The Collective” Our ability to build vast
collective intelligence
and apply it collectives
has made us the top
species
All computers could work
together on a single
problem; whatever is
learned can be instantly
“assimilated” by all
Overall Self Improvement ??? Yes
• Fast computers internetworked
• Decent virtual worlds for AI embodiment
• Halfway-decent robot bodies
• Lots of AI algorithms and representations
• often useful in specialized areas
• often not very scalable on their own
• A basic understanding of human cognitive
architecture
• A cruder but useful understanding of brain
structure and dynamics
• A theoretical understanding of general
intelligence under conditions of massive
computational resources
What We Have Now
Copyright 2018 David Smith All Rights Reserved
22. 4/16/2018
22
Artificial Intelligence in the Movies
Natural Language: Translation
“The flesh is weak, but the spirit
is strong”
Translate to Russian
Translate back to English
“The food was lousy, but the
vodka was great!”
Copyright 2018 David Smith All Rights Reserved
31. 4/16/2018
31
Trust
• Google autonomous car crash.
• Tesla autopilot FATAL car crash.
• Risk assessment for
committing crime racially
discriminates subjects -
COMPASS.
• Predictive policing – racially
profiles the area.
Our fear and mistrust is based
on such failures.
IoT Congress 2017 61
Trust in Military Applications
Signatories:-
Stephen Hawkings, Steve Wozniak, Stuart Russell,
Nils Nilsson and 20,000+
IoT Congress 2017 62
Copyright 2018 David Smith All Rights Reserved
32. 4/16/2018
32
Isaac Asimov’s Three Laws of Robotics
(1940)
First Law: A robot may not injure a human or
through inaction, allow a human to come to harm.
Second Law: A robot must obey the orders given it
by human beings, unless such orders would conflict
with the first law.
Third Law: A robot must protect its own existence,
as long as such protection does not conflict with the
first or second law.
Maybe to AI as well??
Are the 3 Laws the Answer? Extending the
Laws
Zeroth law: A robot may not injure humanity or through
inaction allow humanity to come to harm.
(due to Asimov, Olivaw, and Calvin).
David Langford’s extensions, acknowledging military funding
for robotics:
4. A robot will not harm authorized Government personnel but
will terminate intruders with extreme prejudice.
5. A robot will obey the orders of authorized personnel except
where such orders conflict with the Third Law.
6. A robot will guard its own existence with lethal antipersonnel
weaponry, because a robot is bloody expensive.
Copyright 2018 David Smith All Rights Reserved
33. 4/16/2018
33
• Domains
• Software engineering
• Performance
• Metrics
• Safety
• Usability
• Interoperability
• Security
• Privacy
• Traceability
• Risk Analysis
• Ethics
Pg 65 |
The Area of Standardization
Copyright 2018 David Smith All Rights Reserved
34. 4/16/2018
34
China embracing AI
Imagenet Challenge 2016
• All category winners
were teams from China.
• 33 of the 82 teams from
China.
Revenues from the artificial intelligence (AI) market
worldwide, from 2016 to 2025 (in million U.S. dollars)
Copyright 2018 David Smith All Rights Reserved
36. 4/16/2018
36
World Robot Population
Will They Be Like Us?
72
Like us, AI systems...
...will talk to us in our languages.
...will help us with our problems.
...will have anthropomorphic interfaces.
Unlike us, AI systems...
...will compute and communicate extremely quickly.
...will have bounds for learning and retention of knowledge
that will soon surpass ours.
...might not be well modeled by the psychological models
that work for people.
Copyright 2018 David Smith All Rights Reserved
37. 4/16/2018
37
Atlas, The Next Generation Robot
A new version of Atlas, designed to operate outdoors and inside buildings. It is specialized for mobile
manipulation. It is electrically powered and hydraulically actuated. It uses sensors in its body and legs
to balance and LIDAR and stereo sensors in its head to avoid obstacles, assess the terrain, help with
navigation and manipulate objects. This version of Atlas is about 5' 9" tall (about a head shorter than
the DRC Atlas) and weighs 180 lbs.
Domestic Robots
Copyright 2018 David Smith All Rights Reserved
38. 4/16/2018
38
The Future?
Idea of Artificial Intelligence
is being replaced by Artificial
life, or anything with a form
or body.
The consensus among
scientists is that a
requirement for life is that it
has an embodiment in some
physical form, but this will
change. Programs may not
fit this requirement for life
yet.
Analysis of the Risks
• Mass unemployment?
historical evidence is negative
• Loss of income?
productivity creates wealth, jobs, & ownership
• Idleness & boredom?
the rich are seldom idle or bored
• Loss of control over destiny?
freedom to pursue interests
• Overpowered by superior intelligence?
might bring world peace and economic justice
Copyright 2018 David Smith All Rights Reserved
40. 4/16/2018
40
Paul Allen, Microsoft Co-founder:
“We can see that overall AI-based capabilities haven’t been exponentially increasing
either, at least when measured against the creation of a fully general human
intelligence…individual AI systems…have always remained brittle—their performance
boundaries are rigidly set by their internal assumptions and defining algorithms, they
cannot generalize, and they frequently give nonsensical answers outside of their specific
focus areas.”
… But It Won’t Be Self Aware
Murray Shananhan, Imperial College of London cognitive roboticist:
“Consciousness is certainly a fascinating and important subject—but I don’t believe
consciousness is necessary for human-level artificial intelligence,” he told Gizmodo. “Or,
to be more precise, we use the word consciousness to indicate several psychological and
cognitive attributes, and these come bundled together in humans.”
Peter McIntyre, Future of Humanity Institute at Oxford University:
“By definition, an artificial superintelligence (ASI) is an agent with an intellect that’s much
smarter than the best human brains in practically every relevant field. It will know exactly
what we meant for it to do.”
McIntyre believes an AI will only do what it’s programmed to, but if it becomes smart
enough, it should figure out how this differs from the spirit of the law, or what humans
intended.
McIntyre compares the future plight of humans to that of a mouse. A mouse has a drive
to eat and seek shelter, but this goal often conflicts with humans who want a rodent-free
abode. “Just as we are smart enough to have some understanding of the goals of mice, a
superintelligent system could know what we want, and still be indifferent to that,”.
Richard Loosemore, AI researcher and founder of Surfing Samurai Robots:
Thinks that most AI doomsday scenarios are incoherent and argues that these scenarios
always involve an assumption that the AI is supposed to say “I know that destroying
humanity is the result of a glitch in my design, but I am compelled to do it anyway.”
Loosemore points out that if the AI behaves like this when it thinks about destroying us, it
would have been committing such logical contradictions throughout its life, thus
corrupting its knowledge base and rendering itself too stupid to be harmful.
… And Artificial Super Intelligence Will Make Mistakes
Copyright 2018 David Smith All Rights Reserved
41. 4/16/2018
41
Stuart Armstrong, Future of Humanity Institute at Oxford University:
“Many simple tricks have been proposed that would ‘solve’ the whole AI control problem,”
Examples include programming the ASI in such a way that it wants to please humans, or
that it function merely as a human tool. Alternately, we could integrate a concept, like love
or respect, into its source code. And to prevent it from adopting a hyper-simplistic,
monochromatic view of the world, it could be programmed to appreciate intellectual,
cultural, and social diversity.
But these solutions are either too simple—like trying to fit the entire complexity of human
likes and dislikes into a single glib definition—or they cram all the complexity of human
values into a simple word, phrase, or idea.
“That’s not to say that such simple tricks are useless—many of them suggest good
avenues of investigation, and could contribute to solving the ultimate problem. But we
can’t rely on them without a lot more work developing them and exploring their
implications.”
It Will Be Difficult to Mitigate Those Mistakes
Philosopher Immanuel Kant believed that intelligence strongly correlates with
morality.
David Chalmers, Professor of Philosophy, New York University, and Fellow of the
American Academy of Arts & Sciences:
“If this [Kant’s belief] is right...we can expect an intelligence explosion to lead to a
morality explosion along with it. We can then expect that the resulting [ASI] systems will
be supermoral as well as superintelligent, and so we can presumably expect them to be
benign.”
Stuart Armstrong, Future of Humanity Institute at Oxford University:
“Smart humans who behave immorally tend to cause pain on a much larger scale than
their dumber compatriots,” he said. “Intelligence has just given them the ability to be bad
more intelligently, it hasn’t turned them good.”
“We’d have to be very lucky if our AIs were uniquely gifted to become more moral as they
became smarter,” he said. “Relying on luck is not a great policy for something that could
determine our future.”
Know that ASI Won’t Be Friendly.
Copyright 2018 David Smith All Rights Reserved
42. 4/16/2018
42
However, we won’t be destroyed by ASI.
Eliezer Yudkowsky, Research Fellow, Machine Intelligence Research Institute:
“The AI does not hate you, nor does it love you, but you are made out of atoms which it
can use for something else.”
Peter McIntyre, Future of Humanity Institute at Oxford University:
“An AI might predict, quite correctly, that we don’t want it to maximize the profit of a
particular company at all costs to consumers, the environment, and non-human animals.
It therefore has a strong incentive to ensure that it isn’t interrupted or interfered with,
including being turned off, or having its goals changed, as then those goals would not be
achieved.”
Elon Musk, Founder and CEO of Tesla and SpaceX:
Points out that artificial intelligence could actually be used to control, regulate, and
monitor other AI. Or, it could be imbued with human values, or an overriding imposition to
be friendly to humans.
Super Intelligent Assistants Will Be More Helpful Than Your
Spouse
Know more about your habits
Anticipate your next move
Prepare you for your next event
Provide the right information for events
Communicate your thinking customized for
each recipient
Follow-up on the impact of your decision
Even do the heavy lifting
Copyright 2018 David Smith All Rights Reserved