THE
AGE OF AI - Book Review
The
New Kissinger Doctrine Is a Game Changer
The book The Age of AI
and Our Human Future is a graduate school level text. The Age of AI is the future, and it’s coming
way too fast. The human race has never been more challenged. We are all about
to make some huge decisions.
It is almost a
magisterium for human life in the Fourth Industrial Revolution age. It is
written by thought leaders of the highest-level, each in their respective
fields.
The first author is
Henry Kissinger the former Secretary of State and NSC advisor to two US
presidents, a philosopher and Nobel Peace Prize Laureate. At age 98 he has seen
it all and done it, and remains an international counselor to politicians and
business magnates.
The second author, Eric
Schmidt consolidated Google into the cutting edge technology giant that it is
today. In this role he is a sought out counselor and business mogul.
The third author is
Daniel Huttenlocher – the inaugural Dean of the MIT College of Computing. It is the place where AI is reinvented and
recreated on self-teaching algorithm development and data aggregation from the
global network platforms and the internet that occur 24/7 at a neck breaking
pace.
This compendium though
incomplete, has more authors, contributors and editors. Schuyler Schouten, a
former White House legal counsel.
Meredith Potter is a contributor who augments Kissinger’s intellectual
pursuits she drafted, edited the texts and made the chapters flowing clearly
and seamless. These and other editors
made this textbook intellectually rich, informative, and easy to read.
The Age of AI
introduces the reader to the occurring changes we experienced in our society
today. You are about to encounter many topics that involve the future in its continuing
evolution. Every high school student is adapting to the new classroom
intellectual reality.
Here are two points to
consider.
First, the technology
that this text discusses is not available in your community college courses or
on other educational websites.
Second, whenever you
visit this field knowledge and course material for writing assignments—you may
need more than just 15-20 pages. The
form of the internet search and browsing is different than an online social group
activity. The internet has changed what it means to be educated. Humanity is continuously learning.
This kind of mindset
change is concerning to us and the future winning generation.
The age of artificial
intelligence is here. The book was not written only for geeks and software
engineers. It’s a good core class for every MBA program. It introduces aspects
from the humanities in every chapter. It
has biblical flair to it. It stitches
history and augurs the future, all interwoven in every chapter.
The idea that current
programming code automates some jobs, such as translation, give best travel
directions or loan prediction, has been proven by computer scientists. We
embrace intelligent algorithms to open new opportunities for people with
disabilities. Assistive programs will evolve and eventually replace us as we
know it. Additionally, we will encounter many scenarios where we may not be
sure of what is exactly happening on the other side of the screen.
As the digital world
evolves, so it elevates the level of human intelligence. You feel that I like
this book about Artificial Intelligence and our future.
So let’s get started
with the book.
The Age of AI book has
seven chapters. I studied this book thoroughly while writing the book
review. I review and comment on each
chapter, but not always in their published order. The book is readable even if it’s read in any
chapter order. To be honest I studied
the book chapters at my own order.
Note: To act properly -
whenever I quoted from the book’s text I used quotation marks.
The contents of the
first and second chapters are historical and philosophical in nature.
The second chapter is
titled How We Got Here.
History is an
instructive realm. Particularly fascinating is the evolution of human thought –
who we are and how we got here. The Dark Ages were just that - dark. When the
printing press was invented in the mid-15th century, knowledge, education and
understanding of the world, launched the periods of Enlightenment and the Age
of Reason.
At the 15 to 17th
centuries knowledge became more available and started to circulate in languages
other than Latin. The Christian church
lost the monopoly on understanding the world.
People started to understand Divinity on their own terms. It became a
known fact that there people and large nations in Asia – China, India, and
Americas that think differently. Geography, sea-faring navigation, chemistry
and physics were developed.
The philosophy of the
Age of Enlightenment required freedom of thought. What is real and what is reasonable were
subject of intellectual, philosophical debates.
When knowledge
accumulated and committed to printed books, the first Encyclopediae was assembled
by Denis Diderot during the years 1751-1772.
It had 75,000 entries.
So much knowledge. Neither the State nor the Church liked it. They lost control
over the body of knowledge territorial colonies.
The radical ideas of
personal freedom gave birth to the French Revolution. The Prussian King adopted
the armed reason as a cause of war and started the Seven Years War in Europe.
Conclusion - free
knowledge and encyclopedias have unintended consequences – such as – personal
freedom and sciences.
At the beginning of the
20th century Einstein revolutionized physics. At about the same period Niels Bohr came up
with the Quantum Theory and Werner
Heisenberg with the Uncertainty
Principles. The point is that human
knowledge kept evolving in a faster pace. The World Wars did not stop the
evolution of ideas and knowledge.
During WWII it became
certain that - knowledge is real power! Knowledge is also - physical – It is either
constructive, or destructive.
By end of WWII, atomic
weapons were developed and deployed. The
weaponizing of nuclear energy was followed by peaceful use of nuclear energy
for generating - clean energy… We all
love clean energy… All of us need electricity to power your computers…
The amount of
accumulating knowledge required the perpetual development of powerful and
faster computers. The computers were
mechanical to begin with, and then data was digitized. All manual tasks from half a century ago are
now digitized.
“As
information is contextualized it becomes knowledge. When knowledge compels
convictions it becomes wisdom… As online information has exploded we have
turned to software to help us sort it out, refine it, make assessments based on
patterns and guide us.”
This is where AI enters
our lives.
#
The Third Chapter
titled “From Turing to Today” gets geeky.
Any reader who got this far will understand it.
In 1950 Alan Turing
proposed that in order to measure intelligence we have to assess the external
behavior – the output of the computers.
“...If
a machine operated proficiently that observers could not distinguish its
behavior from human’s, the machine should be labeled intelligent.”
It is known as the
Turing Test.
In May 2020 an AI
platform was released. It was named GPT-3. This name is an acronym for generative pre-trained transformer,
3rd generation.
It can in response to a
prompt generate a human readable, sensible text. How so?
Using texts on the
internet, GPT-3 is trained to generate realistic human text. GPT-3 has been used
to create articles, poetry, stories, news reports and dialogue using just a
small amount of input text that can be used to produce large amounts of quality
copy.
“Given a partial phrase
it can produce possible completions. Given a topic it can produce a paragraph,
and given a question it produces possible answers.”
And more. It can draft
an essay… It can write poetry…
Wow.
All this is done
provided that there is information about the subject matter on the
internet.
So GPT-3 is an information gatherer and a language model. Better yet, GPT-3 trains itself and becomes more efficient
with increasing iterations.
Preceding this
development in 2017, Google’s research team developed an AI chess playing
program branded as Alpha Zero. This
program was given the rules of chess and the chess board game learned to play
chess competing against itself. It
played against itself and while doing so it trained itself to play chess more
skillfully. The chess playing AI
developed strategies unknown to human chess players heretofore.
Yes, we are dealing now
with Machine Learning. The machine
trains itself.
At the heart of the AI
capability to learn are specific algorithms. In Machine Learning, the algorithm is a set
of steps and rules given to an AI program to help it learn on its own how to
improve imprecise results.
The algorithm requires
vast amount of data which it gathers from the internet or from its past
experience.
Well, unless it plays
chess against itself. In that case it
performs millions of chess games and chess maneuvers.
Machine Learning
algorithms technologies are making progress in research labs out there, right
now - as I type and as you read my words.
Better yet, and this is
significant - “The modern AI algorithms
measure the quality of outcomes and provide means for improving those outcomes
enabling them to be more proficiently learned rather than directly specified”.
The AI process has two
phases: first phase is the training. Next phase is the AI quality measurement
and improvement - algorithms evaluate and amend its model to obtain quality
results. The AI reaches conclusions by the model it develops and in response to
reward function that human programmers specified.
This is mind boggling.
The concept and
construct of these algorithms were inspired by the anatomy and physiology of
the human brain.
The most significant progress was made when AI
facilitated transcription and translation of texts. Meaningful interlingual translation was
enabled by mimicking deep neural networks. The generative algorithms became
transformers. And transformers can read
the text from left to right (as in English), or from right to left (as in
Hebrew).
Are you still with
me?
OK. So what really is a
transformer?
A transformer is a deep learning model (algorithm) that
adopts the mechanism of self-attention, differentially weighting the significance of each part
of the input data. It is used primarily in the field of natural language
processing (NLP) and in computer vision (CV)
Transformers do not
necessarily process the data in order. Rather, the attention mechanism provides context for any position in the input
sequence. For example, if the input data is a natural language sentence, the
transformer does not need to process the beginning of the sentence before the
end!
Attention mechanisms
let a model draw from the state at any preceding point along the text. The
attention layer access previous states and weights them according to a learned
measure of relevancy, providing relevant information about far-away items
OK. Let’s parse it in
spoken English.
This Generator that is Pre-trained to create language has this clever Transformer that reads from the start to end, or from the end (of
the sentence) to the beginning and assign weight to the content context of the
sentence in its context.
Here is what happens.
Buckle your seat belt…
The generator can
create text, or images.
Give it an outline and key words and it will write a story.
If a generator can
create a story it can also create a false story – deep fake stories -that can’t
be distinguished from reality. In that
case you need an algorithm that will distinguish and discriminate which ideas
are realistic or fake.
Now we have a struggle
between Generative Adversarial Networks
(GAN).
I call this scenario - The
Wars of The AIs.
This Machine Learning
is now yesterday’s news. Most of it already
happened in 2017.
AI is used for more
than games and story tales. Google’s
Deep Mind project built an AI that temperate the energy by 40% greater
efficiency in Google’s data centers. There the AI performs better than humans. But it takes a human to train the AI to meet
its whims and wishes…
There is a danger in
this clever AI model. Personalizing the output can bias it towards certain
sources and away from other sources. In
plain English we call it censorship… Censorship by AI…
A user does not know how
the AI reached its conclusion and selected its offerings. Even the programmer loses control on the over
the end-results. Yes. AI may make
mistakes but its owners-designers have no idea why.
The AI does not reflect
on its results. It processes the data
that was handed to it and in line with the instructions it received.
AI has no common sense.
AI has no feelings, has no regrets and no grudges.
The continuous
evolution of AI depends on three constraints. Those are the human talents that
write (engineer) the parameters of the AI.
Second are the objectives - of what it is to optimize. And third, AI
processes only what it designed to recognize.
In other words AI does not write its own code as of yet.
Example. Since this book went to print DeepMind, the
subsidiary of Alphabet/Google reported the existence of a new and improved
artificial intelligence algorithm that can perform a wide range of language
tasks—from reading comprehension to answering questions on a broad range of
subjects—better than any existing similar software.
DeepMind’s language
model, which is called Gopher, is
significantly more accurate than these existing ultra-large language models on
many tasks, particularly answering questions about specialized subjects like
science and the humanities, and equal or nearly equal to them in others, such
as logical reasoning and mathematics, according to the data DeepMind published.
Gopher is smaller than
other ultra-large language software. Gopher
has some 280 billion different parameters, or variables that it can tune.
That makes it larger than OpenAI’s GPT-3, which has 175 billion.
AI research,
development, and commercialization happens mostly in the US and China. Unfortunately, one of this book’s short-comings is assessing and scoring
AI in China.
A report in the Hong
Kong daily paper South China Morning Post,
on November 20, 2021 describes advances in a powerful exascale supercomputer in
China that has made a massive increase in its Artificial Intelligence performance.
Aided by a breakthrough
in memory management technology, the New Generation Sunway supercomputer
recorded a 75,839-fold boost in handling data for machine learning. The overall
performance of the computer increased 88 times when processing some of the most
challenging AI-related tasks.
The authors of this
book concede that “the state-funded
Beijing Academy of Sciences has developed a generative language model with 10
times as many weights as GPT-3.” This is still 104 times fewer than the
estimated of the human brain synapses. (There are estimated 125 trillion
synapses in human brain).
A just published
report, from the Harvard Kennedy School of Public Policy concludes that in AI,
China is already a “full-spectrum peer
competitor” of the US, according to the Harvard report, citing a quote from
former Google chief executive Eric Schmidt. It indicated that China is laying
the intellectual groundwork for a generational advantage in AI.
The report indicated
that in deep learning – AI’s hottest subfield – China has six times
more patent publications than the US. It cited an assessment of the Allen
Institute for Artificial Intelligence, which forecast the US to fall to second
place in the top 1 per cent of most-cited AI papers by 2025.
Fact:
In 2019, Chinese universities produced 49,498 PhDs in STEM fields, while U.S.
universities produced 33,759. Based on current enrollment patterns, the report
projects that by 2025 China's yearly STEM PhD graduates (77,179) will nearly
double those in the United States.
That’s the end of
geek-speak. From here on we’ll talk Tik
Tok stuff…
#
Chapter 4 in the book
deals with Global Network Platforms.
Those are the likes of
Google, Meta (Facebook), Amazon (Web Services) and the thousands of Social
Media platforms that utilize AI and algorithms to read our thoughts and sell us
their “stuff” and rake a profit. I bet
that this chapter was authored by Eric Schmidt – the former CEO of Google.
Google, Facebook Amazon
and their likes, won’t be today who they are, without AI and their notorious
algorithms.
Social Media is all
around us – web searches, streaming videos, dating online, ride-sharing, job
hunting, car navigating, political advertising, fund raising, and…
shopping. All make use of AI and
algorithms.
The SM network platforms
deliver valuable daily services and products that provide services to their
users by aggregating the users in large numbers on national and global scale.
The network’s platform utility value grows as its attractiveness grows when
additional users join it.
All these global
networks rely on AI. All the major
global networks are now geopolitically significant. Commercial competition
between global networks affects geopolitical and diplomatic considerations. Materials in a global network may be considered
disinformation in some certain country.
Freedom of expression in one country is defamation in another country.
The common example is
Facebook with its billions of users.
Because of these billions of users and many billions of viewers
“community standards” were created. But
who are the humans that can monitor those community standards?
Here comes the AI to
help. AI does the analysis to determine which content is within the community
standards. From my user experience; can
human supervisors always explain why certain experience was removed? AI has
a mind of its own.
On another front in
2015 Google incorporated AI to the search process with improvement of the
quality and usability of the search results.
That accomplished but at the same time the developers can’t always
explain why a particular page ranks higher than another page… AI has a mind of its own.
One well-guarded,
secretive global network platform is Amazon that turned into Amazon Web
Services. Amazon is influential to its customers because of its clever
algorithms that analyzed my shopping needs and anticipates what I, the shopper,
want to buy, and keeps remind me what else I want to buy. Amazon is not shy to state that” based on my
previous purchases I may be interested in” buying today this other widget or
book… No wonder that Amazon turned also to be a provider of digital web
services such as cloud computing and hosting smaller platforms.
One thing of note. Amazon is not mentioned in this book. Amazon is not listed in the index at the back
of this book. Wonder why?
So this book is not a
perfect compendium….
The author of Chapter
Four on Global Network Platforms proposes a simple explanation of the naked
truth that we are all aware of, as users of Google, Twitter, Facebook, Amazon,
Netflix, Tik Tok and the other popular SM platforms.
It is known as the
positive network effect.
“Positive
network effect occurs for information-exchange activities in which the value
rises in this manner, success tends to produce further success and a greater
likelihood of eventual predominance. People naturally gravitate toward existing
gatherings which lead to larger aggregations of users.
For
network platform relatively unconstrained by borders, this dynamic leads to a
broader, often transnational geographic scope with correspondingly few major
competing services.”
This quote deserves to
be studied, internalize and learned from.
My guess is that this chapter was written by, or mostly written by Eric
Schmidt who led Google for 10 years. He
walked the talk. Traders in international stock exchanges know it best.
On global network
platforms we watch the operation of nonhuman intelligence at global scale.
Presently at the end of
2021, after two years of pandemic caused social distancing AI was placed in
certain networks in a critical and essential resource substitute maintaining
the social glue of society, political governance, and maintenance of business
continuity in terms of remote office work.
I am referring to the rise of Zoom.
Yet the users of Zoom
and similar platforms have no idea where and when AI is managing the video
conferencing processes.
Zoom owners and
operators gained access to the users’ personal information and faces and the
interiors of their kitchens, studios or bedrooms. Same holds for Facebook. Facebook probably holds the facial images of
a billion people on planet Earth. Maybe
more. So the global network platforms
possess and exercise social and political influence. They bear political influence that affects
the most democratic elections.
The fact is that they
muted and banned a sitting US president. The global networks decide what it is
the “community standard”, unless their algorithm decides it faster and ahead of
its human managers. Nonetheless, the global network platforms are social glue.
The following quote is ominously significant,
being a confession by the book chapter author that wrote it.
“AI
operates according to its own processes…
AI
develops its own approaches for fulfilling whatever objective functions were
specified. It produces outcomes and answers that are not characteristically
human and that are largely independent of national or corporate cultures.”
The effects are that
information or recommendations are put out to the info-verse while the
operators and users don’t know nor understand what is occurring in real time.
The common end result is that the owners-operators are frequently embarrassed
by the outcomes and legislatives bodies go after them with reprimands and
occasional legal consequences.
When we talk about
production and distribution of disinformation there is more unsettling
news. The language-generation AI GPT-3
can create synthetic meanings, fake personalities. Those fake users can produce
any fake and any hate contents.
Whille writing this
last sentence I’m reminded of the billion dollars phenomenon known as Harry
Potter series and the question- doubt who wrote all that Hogwarts School of
Witchcraft and Wizardry? Just ponder the
possibilities…
This book chapter
describes in a disappointingly short and fleeting form the growing Tik Tok
phenomenon of vulgarization in social media. Actually TiK Tok is emblematic of
all the potential concerns about AI that I listed above. And more are yet to
come.
Tik Tok was released
for Android and iOS phones in 2017. It
was downloaded over 2 billion times. It
has over 130 million users in the US. It
carries every negative feature that modern day global AI platform can provide.
It offers whimsical, stupid, sometimes salacious short video clips aimed at
youth aged 13 and over. It records and
saves the biometrics – face print and voice print - of its users. It carried
disinformation of any kind and sort.
Governments, both autocratic and democratic, objected to it. Blocked it,
sued it. It is considered a spying agent by proxy of the Chinese
authorities. Tik Tok is all over the
planet and it is not known for sure who controls it, if any. India and America tried to restrict Tik Tok,
Good luck to them.
Global Network
Platforms can change, and do change the norms of societies. The increase consumption, enabled education
through novel modalities, enabled remote work place, changing fashion trends
and electing politician to positions of power.
Global Network
Platforms require a huge amount of original talents. The affluent countries –
rich in resources and rich in talent can afford their own Global Network
Platforms. Most countries can’t afford
it. At the same time “public figures”
are able to use the global platforms to their advantage, their enhanced
visibility and economic exploitation.
#
Chapter Five deals with
country and state security and warfare.
Going back to Proverb
24:6: “With cunning you shall wage war.”
Sun Tzu said: “All
warfare is based on deception.”
Carl von Clausewitz,
the military theorist, said in1832: “Force to counter opposing force, equips
itself with the inventions of art and science.”
Every army has a Corp
of Engineers. The nuclear weapons technology was developed and deployed to win
an end to WWII. Next it was used for civilian purposes and still used in
nuclear powered submarines. Biologic warfare is always forbidden but always
developed in the dark back labs…
Satellites can be
deployed for reconnaissance and military uses. So are rocket sciences.
AI is now integrated in
all these physical armaments.
AI is used to
disseminate disinformation as part of psychological warfare.
AI has the potential to
convert military conflicts beyond restraint. Civilian airplanes already fly and
land with little or no pilot intervention. So can fighter jet during on-air
dogfights. The point is that the race
for AI advantage is already taking place between the U.S., China and Russia.
What really happens
behind the scenes of military R&D remains unknown.
Analysis of the Chinese
Communist Party’s five-year plan by the cyber threat intelligence company
SecAlliance reveals that China is increasingly confident and, no doubt, hostile
in its cyber behavior. This will have implications on the future cyber landscape
and potentially a detrimental effect on industrialized democracies.
Earlier this year,
(2021) the UK Ministry of Defense’s “Digital Strategy for Defense” paper
outlined how the armed forces will access data via a secure, singular, modern
digital backbone. The backbone and a strategy itself are multi-layered. One of
the main aims is to create the ability to exploit vast amounts of data in a
simple way to dominate the battle-space.
On December 17, 2020,
the US Air Force announced the successful flight of an AI algorithm controlled,
known as ARTUµ, (with the pilot) on a U-2 Dragon Lady high-altitude
reconnaissance aircraft. The Secretary said that “Putting AI safely in command
of a US military system for the first-time ushers in a new age of human-machine
teaming and algorithmic competition. Failing to realize AI’s full potential
will mean ceding decision advantage to our adversaries.”
Who are the adversaries
was not specified… We can count on them - the adversaries – that they are also
doing their R&D in due diligence.
For example, The head
of defense contractor manufacturer, Raytheon, Gregory Hays, has said this last
October, that the US is years behind China in its pursuit of hypersonic weapons
which can bob and weave through the atmosphere at more than five times the
speed of sound.
His comments followed
reports that China conducted two hypersonic weapons tests over the summer of
2021, including one of a so-called hypersonic glide vehicle. Launched from a
missile or rocket, the craft separates and zips toward a target while
maneuvering through the atmosphere.
Such weapons can reach
speeds of 22,000 miles per hour, he said. “We have to have automated systems to
defend the homeland, and we are focused on that.”
A missile of this
capability flies at 6 miles per second. Let’s be clear, an AI piloting an
aircraft, is scanning for targets - it follows its own logic - proceeding
faster than the speed of human thought.
Another example, the
Chinese military could be spending as much or even more than the United States
on artificial intelligence (AI), according to a report.
The analysis, by the
Centre for Security and Emerging Technology (CSET) at Georgetown University in
Washington, also found just 22 of the 273 companies known to supply the
People’s Liberation Army (PLA) with AI equipment were not subject to US
Commerce Department restrictions – meaning they may be able to access US
technology and pass it on to the PLA.
Given the secrecy
surrounding the issue, it is difficult to calculate exactly how much each side
spends on AI. Researchers looked at more
than 18,300 publicly available contracts awarded by the PLA and state-owned
defense companies last year. They estimated that Chinese military spending on
AI-related technologies amounted to between US$1.6 billion and $2.7 billion
each year.
Artificial intelligence
can outperform humans in designing futuristic weapons, according to a team of
Chinese naval researchers who say they have developed the world’s smallest yet
most powerful coil-gun.
The prototype weapon
developed by Professor Zhang Xiao and her team at the Naval University of
Engineering in Wuhan has a 12cm (4.5-inch) barrel, about the size of a pistol,
which contains three battery-powered coils that generate an electromagnetic
field.
AI can start with an
imperfect design and make continuous improvements by learning from previous
mistakes, according to Zhang.
The AI gave the human
designers a huge set of optimized data points that nearly doubled the weapon’s
efficiency compared with the US rifle by maximizing the joint performance of
many different components, she said. This resulted in a massive reduction in
the weapon’s size and increased its output energy.
The point here is well
made that AI assisted research and development is used to improve military
armament.
“An
algorithm knows only its instructions and objectives, not morale or doubt… When
two AI weapons systems are deployed against each other, neither side is likely
to have a precise understanding of the results of their interaction will
generate or their collateral effects.”
What clear is, that no
democratic system accountability or verifiable international equilibrium exists
today. There are no known defined AI strategy doctrines. Therefore no
international comparisons are possible.
Before we leave this
subject we must keep in mind that AI generated false content can be used in
psychological warfare. It can generate fake but plausible images of leadership,
photos and videos. It can generate false
speeches.
AI can generate to
produce false images of a country or industry. Both adversarial sides can
employ this process.
One may assume that the
big international powers have each what I call military “Cyber Command.” But I don’t expect these command units
existence to be confirmed in public.
This book does not name
specific countries with such capabilities, like China, North Korea, Iran, UK or
Israel. And there are others.
#
The Fifth Chapter that
deals with National Security and World Order and includes a sub-chapter
discussing the question on how to manage A.I.
The concern stems from
the recognition that most capable countries are now engaged in an AI armament
race. It is recognized that the world faces cyber weapons. AI is weaponized.
Whatever I have quoted here about American and Chinese AI capabilities is
convincing evidence available in the public square. Those capabilities can fall
in the hands of rogue nations and terrorists. Small nations that are not rich
in human, economic, and technological resources can find, recruit and retain
the needed intellectual talent and
resources to create AI systems to fit their national needs.
It is suspected that
lethal autonomous weapon system capabilities that once activated will continue
to act without human intervention or control.
Potential AI managed
weapons systems are like R&D in the field of virology that experimented in
“gain of function”. When virus products escape the lab they incite a global
pandemic. Same applies to cyber weapons.
There is no sure way to supervise or control national AI weapons
development. The best example is use of stock market trading algorithms that
create flash crashes.
In the U.S. the
government does not control AI R&D.
It is done by entrepreneurs, inventors, and universities’ research labs.
It’s dynamic and unrestricted. It is recognized that there are AI-enabled
weapons and AI weapons. The later make
aggressive decisions without human input or human supervision. The challenge
before humanity is to develop and agree upon non-use of AI weapons.
International treaties
regarding control of AI weapons will require international discussions and
negotiations. The AI weapons possessing powers will have to make disclosures of
their current capabilities, will have to outline doctrines of
“within-boundaries-AI R&D”, and preclude attacks on existing nuclear AI
inspection capabilities. Controlling or
abating AI nuclear weapons related capabilities have to occur before
hostilities break.
#
Artificial General
Intelligence (AGI) is the hypothetical ability of an intelligent agent to
understand or learn any intellectual task that a human being can perform.
Stated otherwise, AGI is the representation of generalized human cognitive
abilities in software so that, faced with an unfamiliar task, the AI system
could find a solution.
However we are not
there yet. But the concept is be important when we review .Chapter Six
#
Chapter Six in this
book pivots to the realm of Heideggerian philosophy and social as its title says it all: AI and the Human Identity.
The central question
posed in this chapter is:
“How
will we come to see ourselves and our role in the world? How will we reconcile
AI with concepts like human autonomy and dignity?”
Human societies
traditionally saw themselves as central to their reality. Those are represented
by the traditional or revolutionary leaders, prophets, martyrs, explorers,
inventors, artistic creators, and business moguls.
Thinking used to be
about faith and reason. Human self-understanding is now gradually changing. The
boundaries of reason are stretched and expanded. AI makes better predictions.
AI makes improved products and processes. GPT-3 showed that it can create
better texts and write improved codes for algorithms. Recall the definition of
the upcoming AGI – it is the hypothetical ability of an intelligent agent to
understand or learn any intellectual task that a human being can perform. This
is not science fiction. Human identity is in a state of transformation.
For people who create,
operate and adapt to live with AI, it is going to be an empowering experience.
For people who lack
technical prowess - the passive consumers of AI - it will pose challenges of
understanding why things happen the way they do. The algorithm does not explain why it
happens. Those consumers will lose their
sense of autonomy, sense of fulfillment and sense of identity.
AI generator based
GPT-3 can today write poetry. Will this
poetry be awarded a Nobel Prize in literature?
There will be societal
dislocations.
Every technological
revolution brings with it social dislocations.
When a non-human
intelligence, reasonable, yet inexplicable to humans, makes decisions that
affect humans, then who are the deciding supervisors over the utilization, the
benefits, and the revenue from these AI platforms?
Example.
The AlphaFold platform is barely 4 years old.
AlphaFold is an
artificial intelligence program developed by Google's DeepMind project which
performs predictions of protein structure? The program is designed as a deep
learning system. DeepMind is known to
have trained the program on over 170,000 proteins from a public repository of
protein sequences and structures. The program uses a form of attention network,
a deep learning technique, that focuses on having the AI algorithm identify
critical parts of a larger problem, and then piece it together to obtain the
overall solution. That is the protein’s
3-D structure.
Creating molecules of
proteins became critical to production of mRNA. The beneficiaries of the mRNA
vaccine are billions of people living on Earth today. Yet Pfizer’s and
Moderna’s vaccine brings a bounty of riches transferred from the American tax
payers to the big pharma shareholders. I call it wealth transfer.
Education and lifelong learning receive now a
changed meaning. A new generation of “AI Natives” is growing up with AI
assistants. It started with Alexa and Google Home and it keeps developing into
more complex AI agents. When the
children talk to Siri they don’t know that twenty years ago a cell phone had no
voice assistants. There are anticipated
question about the social adaptation of children who grow up with AI controlled
gadgets. Somewhere else on the globe other children have no 5G phones, no voice
assistants and no AI based educational devices.
There is a growing
chasm around the world between people who are AI literate and other folks who haven’t heard of - AI oblivious. Soon the AI
literate folks will have much greater earning power than the AI ignorant. Civilian occupational fields such as
financial derivative, legal information, library services, meteorology, air
traffic control, car repair, travel guidance, airplane pilots and more have AI
integrated into their working devices.
A new social order will
emerge. The AI-haves already wield
greater earning power than the AI-have-not.
The future will see a
different human identity. Humans will have to strike a partnership with their
AI-integrated machines.
The survival of
Democracy depends on retention of human qualities, maintenance of personal
communication and equal level of reason.
The EU community is
seeking to regulate AI. In China AI is used for surveillance and control the
population.
In America AI is
largely left, so far, to non-governmental organizations. But that will change. The U.S. department of Defense is having its
closely guarded R&D programs. A National AI Initiative Office was
established in accordance with the recently passed National Artificial
Intelligence Initiative Act of 2020. It will be integral to the Federal
Government’s efforts to maintain AI leadership position for many years to come.
#
Ethics
in the Age of AI.
This is Chapter 7 - the last chapter is actually titled: “AI and the
Future.” But who knows what is going to
be the future? Do you know the future?
From the forgoing
chapters’ reviews and discussions it may safely be expected that human ethics
as we know now will also change.
“Technology will
transform knowledge, discovery, healthcare, communication, and individual
thought. Artificial Intelligence is not human. It does not hope, pray or feel.
Nor does it have awareness or reflective capabilities.”
#
The boundaries of
knowledge are expanding. More data and faster analysis processes will open
panoramic landscapes unknown heretofore.
Consequently humanity
will have three choices: to Confine AI, or to partner with AI, or to let AI
loose and defer to it.
Would AI unveil more
than one truth?
We are gradually
entering the phase of the AGI. At this phase AGI will learn and execute a broad
range of tasks. Much like, but more efficiently than humans can perform. Who will control all of that? AI may compel
some of those decisions.
AI is in some instances
was shown to be unpredictable. If that is the case, who will bear
responsibility for the outcomes?
Example.
AI is gradually integrated into nuclear weapons arsenals.
The trigger to the
nuclear weapon is thus carried in a thumb drive.
Read this last sentence
again and let it sink in.
What do you propose to
do?
#
Reviewer’s Conclusion.
The Age of AI is a
scholarly document authored by distinguished architects that formed the reality
in which live. Each is an intellectual visionary in his field.
The book serves readers
from the two ends of the rainbow. At one end are the STEM scholars who get a
refresher view in the history and future of the humanities in the Western
civilization. On the other end are the scholars in the humanities and social
studies who get introduced to the cutting edge technology explained in easy to
understand language.
It is a recommended
text to be required in any college major tracks for students in their senior
year. Be they in philosophy, political science, law, art and design,
engineering, computer sciences, mathematics or accounting.
Humanity as we know it,
is already being divided into two classes.
Those are the AI literate and the AI ignorant classes. The outcome
results in a difference in economic status and standard of living.
The book has its
limitations. Some of its technical
contents have already fallen behind since it went to press.
Certain influential
players in the field of AI are not mentioned.
For example, Amazon Web Services and Apple products. China is a major player in AI R&D and
advances but is barely mentioned. China’ AI accomplishments today, have
possibly exceeded the U.S. in AI technologies.
Last
words – study this book - be AI Literate.
#
Personal
disclosure:
I enlisted the help of
a GPT-3 based AI copy-writer for a few ideas prompts while writing the review.
I purchased on Amazon
my reading and study copy of the book. I
didn’t have an advanced copy of the book, nor did I get to read the book before
its release date on November 3, 2021.
www.mandylender.net
www.mandylender.com
www.visionofhabakkuk.com www.themasterattractor.com
www.attractome.com
#TheAgeofAI
#HenryKissinger #EricSchmidt #DanielHuttenlocher #ArtificialIntelligence
#GPT-3
No comments:
Post a Comment