The document provides an overview of a presentation by John Smart on evolution, development, and the future of networks. It discusses concepts like autopoesis, universal development from outer to inner space, and the "goodness of the universe." The presentation outlines that evolution and development can both be seen in life and the universe, with unpredictable evolutionary processes working with predictable developmental processes to create complexity. It also discusses models of evolutionary development dynamics and examples of evolutionary convergences.
1. The Goodness of the Universe
Evolution, Development, and the Future of Networks
Stepping into the Future
24 Apr 2022 Zoom
John Smart
CEO, Foresight University
johnsmart@gmail.com | @johnmsmart
2. Who Am I?
1. Educator
2. Entrepreneur
3. Foresight Coach
4. Complexity Researcher
5. Husband & Dad
Education:
BA, Business, Haas School, UC Berkeley
MS, Strategic Foresight, School of Technology, U. Houston
MS Eq., Physiology & Medicine, School of Medicine, UC San Diego
Director, EDU Complexity Research Community
JohnMSmart.com
3. Strategic Vision:
What’s Your Theory of Change? Of Adaptiveness?
Good foresight relies on theories and hypotheses of change.
Good theories of change include values, and complex adaptiveness.
I co-founded a community of scholars who study complex systems
from a combination of evolutionary variation, developmental
optimization and adaptive selection approaches. More at:
EvoDevoUniverse.com Georgiev, Smart et al, 2019
4. The Foresight Guide:
ITF: Foresight Practices; BPF: Futures Stories
Nov 2021 Est.: Nov 2022
Free PDFs at: foresightu.com/books
5. Outline
1.Autopoesis (Evo-Devo)
Complex Adaptive Systems
2.Universal Development
Outer to Inner Space
3.Goodness of the Universe
What Keeps Autopoetic Networks Adaptive?
4.Implications and Actions
Seeing and Growing Adaptiveness
“Big Picture, Small Actions”
(Personal, Team, Org)
7. Evolution and Development:
See them in Life, See them in the Universe
Consider two ‘genetically identical’ twins:
Their thumbprints, brain wiring, ideas, behaviors, local processes
and almost all ‘small things’ are unpredictably unique in each twin.
But a small subset of ‘large things’ end up predictably the same,
due to their special initial conditions and constant physical law.
If the Universe is Like Life, unpredictable and creative evolutionary
process works with predictable and convergent developmental
process to create universal complexity.
Uncertainty,
Chance
Predetermination,
Necessity
8. Autopoesis
What is an Autopoetic System?
Greek roots: Auto- (cycle), poesis (creation)
Any physical and informational entity capable of:
A. Replication and self-maintenance (development)
B. Variation and experimentation (evolution)
C. Learning and adaptation (evo-devo)
All autopoetic entities encode (“entrain”) physical and informational
models that allow them to continue to exist in their particular
environment. Those models help them outcompete other uses of resources
in those environments. The greater their autopoetic complexity, the more
generally adaptive knowledge (applicable in all environments) is entrained.
Some Autopoetic Systems:
Prebiotic chemicals, genes, cells, organisms, species
Facilitated/Dependent (not obviously self-replicating) Autopoesis:
Suns, Viruses, Brains, Language, Ideas, Institutions, Technologies.
Campbell 2021
9. Evolutionary Development (Evo-Devo) Dynamics:
The ‘Left and Right Hands’ of Universal Change
“Experimentation”
Stochastic Search
Strange Attractors
Radiation
Development
‘Right Hand’ of Change
Evolution
‘Left Hand’ of Change
Well-Explored Phase Space ‘Optimization’
New Computational Phase Space ‘Opening’
“Convergent Unification”
Environmental Optimization
Standard Attractors
Hierarchy
“Natural Selection”
Requisite Variety
Mixed Attractors
Adaptation
Evo Devo
(Intersection)
Carroll 2005
The Evo-Devo Process Philosophy
(Complex Systems Framework)
10. Three processes (telos) can be
clearly seen in five hierarchical
(developmental) systems:
Technological Systems
Societal Systems
Biological Systems
Chemical Systems
Physical Systems
Evo-Devo Pyramid:
Three Basic Processes in All Complex Systems
Evo Devo Universe?, J. Smart, In: Cosmos & Culture, Steve Dick (ed.), 2009
“Uncertainty” “Determinism”
12. Evo-Devo (Autopoetic) Systems:
Experimentation + Selection + Stabilization and Replication
Transitions from:
“Quantum Darwinism” to relativistic physics (Zurek 2005)
Stellar nucleosynthesis to life cycle (Clayton 1968)
Molecular evolution to biogenesis (Smith and Morowitz 2006)
Multicellular evolution to tissue types (Newman and Bhat 2008)
“Neural Darwinism” to brain development (Edelman 1989)
Intracognitive selectionism to conscious thinking (Calvin 1985)
‘Memetic’ social selection to moral universals (Wright 2000)
‘Technetic’ selection to technological archetypes (Kelly 2010)
Biomimetic computation to machine intelligence (Doursat 2008)
Cosmological natural selection to universe development (Smolin 1992)
EvoDevoUniverse.com
13. The “95% vs. 5%” (95/5) Rule:
Observed Evolution vs. Development
▪ Almost all genes in an organism (eg, 95% of Dictyostelium DNA)
change over macroevolutionary time to create evolutionary variety vs. a
highly conserved subset (5%) which form the developmental toolkit.
▪ Almost all cells in an organism compete for their location. A special
few are fated in advance to particular locations early in development.
▪ Almost all ideas and actions in an organism are experiments. A
special few become stable strategies, across many environments.
▪ Almost all technology products and services are evolutionary
experiments. A few are developments destined to be the next big thing.
Almost all (roughly 95%) of the observed physical
and informational processes that create or control a
complex system are bottom-up, local, divergent
stochastic evolutionary processes.
A critical subset (~5%) of observed processes are
top-down, global, convergent, statistically predictable
developmental processes. Examples: 5% Devo
95% Evo
14. Evo-Devo Foresight:
“Three Ps” of Future Thinking, Three Management Groups
Leadership, Strategy, Analysis, Planning
Innovation, Ideation, KM
Design, Entrepreneurship
Forecasting, Financial,
Risk Mgmt, Law & Security
Seeing
“Trees”
Future Oriented
Seeing
“Funnels”
Past Oriented
Seeing
“Landscapes”
Present Oriented
Alvin & Heidi Toffler 1970
20. Computational Incompleteness
• GAIs Will Never Be Godlike (Strongly Omniscient or Omnipotent)
• They’ll Remain Like Us. Inductive. Emotional. Belief-Based Entities.
• They’ll Thrive on Differences, and a Diverse Network of Crowds.
• Kurzweil is Likely Right that They’ll Have Religions, Too.
• How much Encoded Wisdom Must the Universe Accumulate
Before You’d Consider it a Higher Power?
21. The Evo-Devo Framework is a Process Philosophy:
It Contains Many Related But Unique and Falsifiable Models
The Evo-Devo Framework is not a single theory. It is a process
philosophy with similarities to other useful process philosophies
(Alfred Whitehead, Charles Pearce). It is a systems “theory,” a
framework for analyzing physical and informational dynamics
(processes) in complex adaptive systems.
It is an epistemic framework, not a “theory of everything”. It
proposes that there will never be any such descriptive theory, but a
vast number of contextually useful and incomplete “theories of
special things”, like the large and inelegant set of variables in the
seed (initial parameters), the replicating organism, and the selective
environment that guide adaptiveness in living systems.
So how do we best test its generality and validity?
It can be incrementally supported or falsified as we develop and
compare our models of autopoesis in many specific contexts.
22. A Large Set of Evo-Devo Models Deserve Evaluation
Many specific and falsifiable models fit either largely or entirely under
the autopoetic evo-devo framework. Here are a few worth testing:
• Terry Deacon’s Autogenic Networks (Neither Ghost Nor Machine, 2017)
• John Campbell’s Bayesian Inference (The Knowing Universe, 2021)
• Karl Friston’s Free Energy Principle (Active Inference, 2022)
• Peter Corning’s Control Information and Synergism (Complexity)
• Wojciech Zurek’s Quantum Darwinism (Physics)
• Addy Pross’s Dynamic Kinetic Stability (Chemistry)
• Gerald Edelman’s Neural Selectionism (Neuroscience)
and more speculative models like:
• Richard Dawkins’ Memetics and Susan Blackmore’s Technetics
• Danko Nikolic’s Practopoetic Networks (AI design)
• The Natural Intelligence and Security Hypotheses
• Universal Autopoetic Values Models (The IES Goals, etc.)
• John Smart’s Transcension Hypothesis
Science must evaluate and seek to falsify any evo-devo congruent
models relevant to the adaptive context under investigation.
24. A U-Shaped Curve of Change:
Inner Space to Outer Space Back to Inner Space Again
Big Bang Singularity
100,000 yrs ago: H. sap. sap.
1B yrs: Protogalaxies 8B yrs: Earth
400,000 yrs: Matter
70 yrs ago: Machina silico
50 yrs: Scalar Field Scaffolds
Developmental Singularity?
25. Evolution Versus Development:
‘Search Landscapes’ Versus ‘Portal Pathways’
Crutchfield, J.P. 2001. When Evolution is Revolution: Origins of Innovation. In: Crutchfield & Schuster
(eds.), Evolutionary Dynamics: Exploring the Interplay of Selection, Neutrality, Accident and Function.
Research Questions:
• Are portal paths that lead to greater general adaptiveness plentiful or rare?
• Are such portals sequence-dependent or randomly traversible?
• Are such portals convergent, divergent, or non-vergent (as depicted here)?
Evolutionary Convergences (Portals, Bottlenecks, Attractors)
Must Exist if Universal Development Occurs
26. Evolutionary Convergences
(Developmental Attractors): A Few Examples
Evolutionary convergences are developmental attractors in a
hierarchical physical-computational universe:
Organic (carbon) chemistry (vs. silicon, boron, etc.)
Amino acids, purines, pyrimidines, lipids as cell precursors
RNA as enzyme and code for protein architectures (Woese)
Dynamical pattern modules in multicellularity (Newman 2008)
Antifreeze molecules in northern and southern polar fish
Placental vs. marsupial mice, moles, rabbits, wolves, tigers, etc.
Eyes, body plans, limbs, joints, wings, fins, emotions (Morris)
Bilateral symmetry, binocular vision, tetrapod form
Prehensile limbs, opposable thumbs, anthropoids (Russell)
Mimicry memetics (languages) gestural, behavioral, oral, written
Neolithic+ tools (rock, club, spear, lever, rope, wheel, pulley)
Internal combustion engine, metallurgy, chemistry, electronics,
Math, science, computers, internet, smartphones, cloud, AI…
Kelly 2010
Conway-Morris 2003
27. Evolutionary Convergences: Why They Occur
Some convergences greatly advance individual and cultural
information processing and adaptation in a broad range of
environments, for the first species that acquire them.
Eyes, jointed limbs, body plans, emotions, imagination, language,
opposable thumbs, tool use, etc.
The streamlined shape of fish fins, first created as an evolutionary
morphological experiment, must persist in the genes of all organisms
seeking to move rapidly in water on all Earth-like planets, as a universal
developmental constraint imposed by our particular universe’s physics.
Such advances are ‘evolutionary ratchets’ (function randomly acquired
but statistically irreversible once acquired, in a broad range of
environments), a type of developmental optima (for a given level of
environmental complexity) in all universes of our type.
28. A Portal Pathway for Chemical Autopoesis:
Carbon and Third Generation Stars
Carbon is the only way forward to complex (living) chemistry. Boron and silicon, and non-liquid phase
environments, are no longer considered viable to form high-complexity autocatalytic cycles.
Four of the six most common elements (CHNOPS) in life chemistry, and both of the great oxidizers
(oxygen and sulfur) only become plentiful in small, third gen (Pop I) stars, like our Sun.
Genesis of Chemical Elements
29. A Portal Pathway for Life (Biological Autopoesis):
RNA, Lipids, Proteins, and Sugars
• Lipids and RNA in particular may be the only way from organic chemistry to cells.
• RNA, lipids/cell membranes, and protein precursors (amino acids) all form spontaneously in
Earth’s chemistry (and precursors form on meteorites).
• Nucleobases (AGCT/U) form from cyanide, acetylene and water.
• Sugars form from alkali and formaldehyde
• Phosphates are released from schreibersite in meteorites (“solar
system assist”), and from (modern) volcanic vents.
• Sutherland et. al., by mixing sugar and nucleobase precursors and
phosphate got 2-aminooxazole (partial sugar, partial nucleobase)
• Exposure to intense solar UV in shallow water (“solar system assist”)
destroys the incorrect forms of nucleobases, leaving behind C and U.
• RNA is today the only known heteropolymer (of 10M chem. species) that can both reproduce
and catalyze 3D construction (protein, via ribozymes)
• RNA later learned to store itself more permanently (stably) as DNA
(RNA World Hypothesis), DNA may be path dependent on RNA.
Alonzo Ricardo and Jack Szostak. Life on Earth, Scientific American, Sept 2009.
Matthew W. Powner, Beatrice Gerland & John D. Sutherland. Synthesis of activated pyrimidine ribonucleotides in
prebiotically plausible conditions, Nature V. 460 May 13, 2009.
32. Free Energy Density (Phi, Φ) of Autopoetic Systems
Traces out a Universal Hyperbolic Curve
Free energy flow density in hierarchically emergent
autopoetic systems (note some, like Tech, Culture,
Brains, Galaxies & LSS are facilitiated autopoesis).
Energy Flow Density (Φ)
Substrate (ergs/sec/gm)
Weak AI of Early 21st C 10^12+
Pentium II of the 1990's (10^11)
Intel 8080 of the 1970's 10^10
Modern Engines 10^5 to 10^8
Culture (human) 500,000 (10^5)
Brains (human) 150,000 (10^5)
Animals (human body) 20,000 (10^4)
Ecosystems 900
Planets (Early) 75
Stars 2
Galaxies 0.5
Large Scale Structure 0.05?
Chaisson 2001
33. Annual Growth Rates of Familiar, Human Scale Systems
2.5% US Economy, Many Developed Nations GDP
3.7% Real Global Economic Growth (Brynjolfsson, GDP-B)
9% China’s Economy, Some Emerging Markets
15% IoT Growth, Global eCommerce, Top Chinese Cities GDP
20% Most Technology and eCommerce Markets
25% Health Care Analytics, India’s Organic Food Market
The Digital World is where things get unfamiliar
40% Twice the Computing ability per Dollar every 2 yrs (“Moore’s law”)
55% Creation and Sharing of Digital Info (“Tagging the World”)
60% of the world’s data was created in the last 24 months.
10X more data every five years.5,200 GB for every person on Earth.
210% Deep Learning Training Data and Parameter Sets
Data Lakes and Parameters now growing 10X Every 2 Years
The Quantum World gets vastly more unfamiliar
!?! Q. computers grow on double exponential rate. (“Neven’s law”)
221
, 222
, 223
, 224
, 225
(4, 16, 256, 65K, 4.3 Billion), etc.
Rates of Accelerating Change
35. Infotech/IT/Simulation - Virtual Inner Space - Evolution
“As Intelligence Rises, Thinking Becomes More Adaptive Than Acting”
Adult humans no longer act in novel ways, they think in novel ways.
Simulations allow “ephemeralization” (far less mass/energy per action)
Rise of scientific simulations. IPCC. NASA Solar System Simulator
Telepresence outcompetes traveling for perception
Telerobotics/haptics outcompetes traveling for action
Google maps, sensors, geoweb, parallelized GPUs: visual cortex for the web.
Machine sim data doubles every 2 years. Human sims grow far slower.
Nanotech/Engrg - Physical Inner Space - Development
“There Are Orders of Magnitude More Performance at the Bottom.”
Fission 1,000X more E density than chem. Fusion 1,000X more again than fission
Fuel cells can store 100,000X more E/mass than chemical batteries
Synthetic catalysts increase reaction speeds and yields 1,000-1,000,000X
Programmable synapses use 10^6 less energy per comp. than neurons
Photonic crystal lasers 10^6 more energy efficient than other microlasers
Single step efficiency jumps in macro (human) space are always far less
The Race to Inner Space:
Civilization’s Hidden Strategic Objectives
The Future of Scientific Simulations: From A-Life to Artificial Cosmogenesis, C. Vidal, 2008
36. Evolutionary Trends:
Diversity
Specialization and Individuation
Innovation and Freedom
Collective Intelligence
Fair Competition
Technological Unemployment
Global Innovation
Developmental Trends:
Social Democratic Capitalism (UBI)
Morality and Empathy
Evidence-Based Behavior
Tech Autonomy, Natural Intelligence
Human-Machine Merger
Transparency and Security
Global Sustainability
Engineering and Infotech Accelerations
Are Driving Social, Economic, and Political Change
GDP per capita in Western
Europe, 1000-1999 AD.
We live in exponential times.
We may all end up like Sweden, more or less.
Inglehart & Welzel, WorldValuesSurvey.org
37. Densification and Dematerialization:
Why Acceleration Happens
“Densification” “Dematerialization”
D&D is: Densification (miniaturization, urbanization, automation), and
Dematerialization (substituting info and computing for “things”)
D&D drives our ever faster, wealthier, smarter, and more resilient world.
The most densified and dematerialized systems on Earth? Human brains.
Our world is gaining a “global brain”. New digital networks always emerging.
How we guide this process, and whether we empower people, is up to us.
“Density Beats Sparsity” “Bits Beat Atoms”
38. 75% of Americans (250 million) now
live in just 3% of US Land that is Urban.
IT, Pharma, Biotech are our fastest growing Industries.
We are an Urban, Corporate, and Nano Economy.
US uses less of most physical resources per
capita than 1980s, and has peaked pollution
& CO2, even as our GDP accelerates.
We are an Intangibles Economy.
Localization, Miniaturization, Digitization, Simulation:
D&D Drivers of Physical and Virtual Inner Space
Localization
(Denser and More
Powerful Systems)
Miniaturization
(Nano Replaces
Macro Systems)
Digitization
(Bits Replace
Atoms)
Simulation
(Software Replaces
Actions)
in Tangible Systems in Intangible Systems
Densification Dematerialization
39. Look to Barrow’s and Sagan’s Scales, Not Kardashev’s
Sagan’s Information Mastery (Dematerialization) Scale
Based on the total bits of useful information available to a civilization.
Starts at 10⁶ bits, assigned this the letter A. 10⁷:B, 10⁸:C, … ,10³¹:Z.
In 1973, humanity accessed 10¹³ bits, Level H. By 2020, we were at Q.
This scale proposed to measure “unique” information, but without specifying
how. Thus it presently includes no measure of knowledge or general
adaptedness. Also, 10³¹ bits is just 1Kg of Black Hole Computronium.
In reality, we’re going to need a lot more letters.
Barrow’s Civilization Development (Densification) Scale
Impossibility: The Science of Limits and the Limits of Science, John Barrow, 1998, p. 133.
The Cosmic Connection: An Extraterrestrial Perspective, Carl Sagan, 1973.
41. Smolin’s Answer:
Autopoetic Systems (Cosmological Natural Selection)
At least 8 of the 20+ standard model
parameters appear to be fine tuned for:
– fecund black hole production
– multi-billion year old Universes
(capable of complex, long-lived Life)
Smolin 1996
43. Smart’s Answer:
Merger Devices for Civilizations
Smart 2008
Dark Energy Divides Our Cosmos into “Follicles”.
In Each, Local Galaxies, and their Civilizations
Can Merge in Black Hole Time (Instantaneously)
Active Merger May Also Occur
44. The Transcension Hypothesis:
Are Black Hole Like Domains Our Developmental Destiny?
The Transcension Hypothesis: Do Advanced Civilizations Invariably Leave Our Universe?,
John Smart, Acta Astronautica, 11 Dec 2011.
Key Assumptions of the Transcension Hypothesis, John Smart, EverSmarterWorld, 2016.
Earth is a cosmologically
dense emergent intelligence.
Is our future even denser and more risk immune?
Black holes may be a developmental destiny and standard attractor for all
advanced intelligence, as they appear to some physicists to be ideal
computing, learning, energy harvesting, forward time travel, civilization
merger, natural selection, and universe replication devices.
45. Both Reproduction and Birth are Energetically
Accelerative in Complex Living Systems
• Note the energy flow trends
in each phase of the life
cycle (birth, growth,
reproduction, senescence) of
a developing organism.
• The reproduction, birth and
early growth phases are
each accelerative.
• Stellar autopoesis, via
supernovas, has a similar but
simpler energetics profile.
• Does universal complexity
accelerate because it is
engaged in a process of
universal reproduction?
Salthe 1990
46. Our Childproof Universe
Vast Diversity and Isotropy of Intelligence (Massive Parallelism)
Gamma Ray Bursts Too Isolated to Threaten Neighboring Life.
Gas Giants (Jupiter, Saturn) Remove Planet-Killing Meteorites Early in
Earthlike’s Life Cycle.
Gaia Theory (Biogeoclimatic Homeostasis). Climate Change is Bad for
People, Good for Life.
No Theoretical Limit on Thermonuclear Device Yields, But a Low
Practical Limit. No Doomsday Device.
Nuclear Winter is Opposed By a Homeostatic Climatosphere
Large Antimatter Bombs Energetically Impossible To Build
Pandemics Mutate and Breed Immunity (no Species-Killers)
Humans and GAI May Both Require Natural Intelligence (Ethics,
Empathy, Emotions, Uncertainty, Diversity) and Natural Security
Extensive Statistical Guardrails Seem to Protect Complex Intelligence.
We Face a Panoply of Nonexistential Risks.
47. Goodness of the Universe
What Keeps Autopoetic Networks Adaptive?
49. • Empathy
(connection, love, understanding, compassion)
• Ethics
(fairness, justice, responsibility, equity)
• Empowerment
(individual and collective rights, abilities, wealth)
• Evidence-seeking
(science/discovery, inquiry, data, rationality)
• Expression
(freedom/creativity, beauty, experimentation, fun)
* “Plato’s Triad” of Societal Values “The Good, the True, and the Beautiful”.
Five E’s of a Good Society*
50. What is a Human? A Curious Blend of Head, Hand, and Heart
Oldowan axes mass-produced for
cooperative and competitive use by
Homo habilis ~2 million years ago.
Wrangham 2019
We Cooperate First, and Compete Second (We are a “Coopetitive” Species)
52. Catalytic Catastrophes: Speed and Strength from Adversity
In Diverse, Learning-Capable Networks, Catastrophes are Catalytic.
Disruption For Some is Always Making Someone Else More Adaptive
Wood 1981/2022 Oct 2020
Ice Age Toolkits and Culture
Mammalian Complexity
53. Catalytic Catastrophes and Immunity
• Many catastrophes in Earth’s history, like the Permian Extinction,
K-T meteorite, Paleolithic Ice Ages, and our History of Warfare,
while painful, have also accelerated complex adaptiveness.
• Complex networks learn from, and even thrive with, catastrophe.
This is why accelerating change is so smooth on long timescales.
• We prevent catastrophe if we can, and learn from it when we can’t.
• To protect society in our technologically advanced future, we must
build a planetary immune system that learns rapidly from threat.
• Both bottom-up (souveillance) and top-down (surveillance)
transparency must be created, and our privacy protected.
• A healthy living system is transparent to a trusted immune system
and compartmentalized (private) to everyone else.
• In late 21C society, privacy, compartments, and secrets
(national, trade, personal) must still exist, yet we must also
have global immune systems (IoT, cameras, robots, AI) that go
everywhere, and uncover the bad actors when crime occurs.
Brin 1998
54. Global Economic Growth had
a State Switch Circa 1880.
Tangibles Drove the
Industrial Revolution.
Intangibles Drive the
Intelligence Revolution
(“4th Industrial Rev.”).
Half of all the economic wealth in human history
was created in the last 20 years (1998-2018).
Our next wealth doubling will be even shorter (2019-2035).
“Bits Now Winning Over Atoms” – Kartik Gada
55. Our Accelerating Economy
• Tech and data productivity, are steeply deflationary. Together they drive ~70% of economic
growth. Another ~20% is labor productivity, ~10% is finance productivity. (Robert Solow).
• America’s growth in GDP per capita, and our inequality, have led the world since 1930.
They’ll keep doing so, if useful data, digital goods, automation, and AI keep growing.
56. What Caused Our 1800s “State Switch”?
New Beliefs, Technology, and Networks
New Beliefs (Enlightenment)
▪ The rise of science (theory and experiment) in Europe in the 1600s.
▪ New ethics (values), empathy (feelings), and beliefs (reform Christianity).
New Technology (Industrial and Information Revolutions)
▪ Far more powerful, stronger, and faster machines.
▪ Cheaper energy, transportation, production, and resources.
▪ Growing value of information and intelligence (bits over atoms)
New Networks (forms of Cooperation and Competition)
▪ Property rights, corporate charters, patents, capital, rule of law (economic networks).
▪ The reemergence of democracy in the US and Europe (political networks).
▪ Education (often via mass military conscripting) and middle class expansion.
More On Our New Beliefs: After the Enlightenment, we began to talk less of
conquerors and kings, and more about good leaders, entrepreneurs, scientists,
inventors, educators, artists, and activists. We began to believe we could make
progress, here on Earth. We gained new ethics of thrift, persistence, innovation,
competition, and new empathy and respect for the freedom and worth of others.
Beliefs Matter!
History of the Idea of Progress, Nisbet 1980
The Bourgeois Virtues, Dierdre McCloskey, 2006
57. Three Questions for Managing Accelerating Change
1. Beliefs (Head). Are yours evidence-based?
Adaptive? Do you see progress? Do you take
responsibility? Have a vision?
2. Technology (Hand). How effectively do you use
tech? Are you benefiting or being disrupted by it?
3. Networks (Heart). How well are your caring for
and using your networks? Do empathy and
ethics come first? Are you cooperating primarily,
and competing secondarily, as best you can?
Fortunately our attitudes and beliefs, including our ethics and
empathy for others in our networks, are the life variables that
we can most freely choose and control. They are the greatest
determinant of our lifelong happiness as well.
Frankl 1946
58. Natural Intelligence (NI) Thesis:
AI Minds Will Become Increasingly Like Ours
Must AI Become NI? Is there “No Other Easy Path”?
There is a popular thesis that AI could easily become an “Alien Intelligence”. I disagree. I
think life, with its vast evolutionary search capacity and its miraculous error-correcting
developmental capacity, has found the only accessible path, in noncosmologic time. What
is more amazing than biological evolution and development?
Nature prefers to copy and vary. It does not recreate complexity from scratch.
I expect we’ll be forced to copy and vary with AI. We’ll engage in deep neuromimicry and
biomimicry, guided by our advancing neuroscience and growing understanding of
biological evolutionary development, under continuous selection and adversity.
If NI is true, an increasingly independent autopoetic path to general AI will outcompete our
current “rational designs” (actually, very slowly facilitated autopoesis), as it can be far more
generative of evolutionary variety, and far more constrained by development, network
dynamics, and selection, than any human-designed performance and alignment.
If NI is true, our Best AIs will Develop in a Collective Network, Policing Each Other (think of
GANs + Logic, Emotions, and Ethics). They will become Coopetitive. They’ll have not only
Pattern Recognizing, but Emotions, Logic, Ethics, Empathy and Evo-Devo Dynamics.
59. Deep Learning Has a Little Neuromimicry, But No Biomimicry
A Mouse, A Laser Beam, A Manipulated Memory
Liu and Ramirez, TEDxBoston 2013
Deep Learning Mimics Human Sensory Cortex
Yamins & DiCarlo, Nature Neuroscience, Mar 2016
61. Deep Learners are a Very Long Way from GAI
Today’s Deep Learners have:
• No Compositional Logic
• No Commonsense Reasoning
• No Self-, Other- and World-Models
• No Emotions and Empathy
• No Network Ethics
• No Artificial Development
• No Self-Reproduction under Selection.
In the NI Hypothesis, all this and more will be
required to get, in adaptive time, to General AI.
Today’s deep learners won’t be allowed to
drive cars at scale, in my view. No commonsense
reasoning, no world model. Endless edge cases.
Marcus & Davis 2020
62. Naturally Intelligent Machines: From Teaming to Merger
Future NIs, Teamed with Humans (Personal AIs),
Will Do Lifelong Uploading and Mind Melds (Merger),
In a Process of Convergent Evolution (Development).
Network Interdependence. Not “Alien Intelligence”.
Spock mind melding with a computer, Nomad
The Changeling, Star Trek: TOS, S02E03, 1967.
Swarm cognition.
Seeley 2010
63. The Singularity is Not Near
Some say “The Singularity (General AI) is Near”.
Deep Learning has been on a hype peak since 2017.
In reality, DL today is First-gen Neural Nets doing mainly
dimensional reduction (finding patterns in big data sets),
plus a little reinforcement learning, adversarial networks,
sentiment, and other weakly bio-inspired approaches.
When can we expect GAI? Circa 2080s, in my view.
Underappreciated evo-devo complexity (mainly), and
Regulation and testing as it gets creepy (secondarily)
will delay it well past the common “AI expert” polls.
Kurzweil 2005
64. How will we make deep learners safe?
Autopoesis, Artificial Selection and Natural Security.
From domesticated animals to domesticated robots/agents.
Sagarin (Ed.), 2008
65. Natural Security Thesis:
Alignment and Security Will Require Deep Biomimicry
Is there any alignment and defensive strategy better than
network empathy, ethics, and intelligent immunity?
If NS is true, we’ll use the same techniques to goal-align and secure AI and robots that
we presently use for domestic animals and people. We’ll Build Statistical Empathy,
Ethics, and Security based on Testing, Selection, and Differential Reproduction.
We’ll Watch For: Bad Algorithms, Bad Data, Bad Training
We’ll Replace those with: Validated Algorithms, Better Data, Better Training
• Safety in Testing (Audited Behavior)
• Safety via Selection (Differential Reproduction)
• Safety in Numbers (Trustable Swarms)
• Safety in the Network (Average Behavior)
There will always be AI Rulebreakers, Fanatics, and Sociopaths, on the Network Edge.
We and our AIs will need to defend against them statistically, as we always have.
We’ll need Labrador AIs to fight the Coyotes and Raptor AIs to fight the Rogue Raptors.
66. Natural Security:
Moving from Physics Thinking to Biology Thinking
The more complex & connected the world gets, the less useful physics
thinking becomes. Life employs the most complex and adaptive defense
networks, by far. Increasingly, we must learn from nature to defend.
Physics Thinking: Mathematical, Logical, Precise, Engineered, General,
Optimized Systems and Outcomes (our “5%, future”)
Biology Thinking: Nonlinear, Connectionist, Imprecise, Grown, Evolved,
Trained, Selected, Adapted, Specific, “Gardened” Systems and Outcomes
(our “95% future”) (“95/5 Rule”)
Today, the strong bias of leaders is to physics thinking. That must change.
Arbesman 2016
Sam Arbesman,
Scientist-in-Residence,
Lux Capital (VC Firm)
Rafe Sagarin (1971-2015).
Late Founder and Director,
Ctr for Natural Security, U of AZ
Sagarin, 2012
67. Natural Security: What Can We
Learn from Biological Immunity?
Biology Creates Immunity by:
1. Decentralization / Crowd Empowerment / Swarm Defense
2. Transparency (Only 5% Top-Down, 95% Bottom-Up)
3. Diversity of Local Responses / Experiments / Trial & Error
4. Local Memory / Permanent & Accurate (B-Cells, 12X)
5. Local Feedback / Rapid Problem Reporting (Dendritic Cells)
6. Redundancy of Critical Systems / Fault-Tolerant Networks
7. Redundant and Competitive Defense Networks
(Complement, T, B, NK, Neutrophils, Macrophages…)
8. Stress Testing / Penetration Testing / Antifragility
9. Intelligence / Integration, Simulation and Prediction
10. Scale Invariance (As Fast in a Mouse as in an Elephant)
This is an Incomplete list.
Can We Measure and Improve Immunity Capacity?
Can We Ensure a Rapid, Accurate Responses, at Every Scale?
68. Natural Security:
Good Networks Always Win (Survive, Improve, Adapt)
The Best Networks:
• Empower Individuals Over Leaders (20:1 Decentralized:Centralized)
• Champion Diversity (Specialization), Comms, & Trade (Interdependence)
• Experiment (Innov) & Learn from Failure (Catalytic Catastrophes)
• Evolve Fair Rules, Transparency, and Trust
• Grow Redundancy & Resiliency (Fault-Tolerance, Institutional Memory)
• Get Stronger Under Stress (Power, Antifragility, Immunity)
• Build Collective, Collaborative Intelligence (“Wisdom of Crowds”)
Q: Are You Tending Your Networks? (Personal, Org, Soc, Tech)
Q: Will they Be Stronger and Smarter When You’re Gone?
Phylogenetic Networks Social Networks Knowledge Networks (“Graphs”)
Neural Networks
(Biological, Artificial)
70. A Dozen Mistaken Transhumanist Assumptions
1. “Evolution is a Blind Architect.”
2. “We Live in a Deterministic Universe.”
3. “GAIs will be Godlike. They’ll be able to deeply optimize their futures.”
4. “We must rationally engineer and goal-align advanced AIs.”
5. “Future AIs can easily become misaligned with humanity.”
6. “The far GAI future is an “event horizon” (conceptual singularity).”
7. “Most aspects of life could change rapidly and radically after GAI.”
8. “We are on the cusp of radically enhancing our biology.”
9. “Much of today’s regulation is “antiprogress.”
10. “Humanity is surrounded by existential risks.”
11. “Transhumanists are change- and future-oriented, humanists are not.”
12. “Our future is fully ours to choose, and freer than we can imagine.”
71. “Evolution is a Blind Architect”
Corollary: Science and rationality allow us to “escape evolution.”
Reality: Both biological evolution and rationally-aided human science and
action are autopoetic (evo-devo) processes, operating in different complex
systems. Both are computationally incomplete. Both are always guessing,
feeling, generating and testing models and beliefs about the future. Both
encode a deep autopoetic network intelligence.
Implications/Actions:
The better we understand the adaptiveness and intelligence of nature’s
autopoetic networks, the better we can live, build, and act.
Development must be seen on par with evolutionary (random, contingent,
creative) change. There is nothing more amazing in our universe than
biological development. We must harness it to get to GAI (NI).
Seeing Evo-Devo processes in all adaptive systems, including our
Universe, is one key to the Extended Evolutionary Synthesis we need.
72. “We Live in a Deterministic Universe”
Corollary: Deduction, rationality, and causality are always the most
valuable ways to build an epistemology.
Reality: Our Universe is only partly deterministic (developmental). Quantum
mechanics itself tells us this. Reality is both unpredictable and predictable
(evo-devo). Creative and conservative. We use induction ~20X more than
we use deduction (95/5 Rule). Causality can only partly describe it, and math
is more invention than discovery. We always feel our way to the future.
Implications/Actions:
We must blend induction, deduction, and abduction (probabilistic models,
analogy) to understand the world, and most of the time, use induction.
If ours is an evo-devo universe, causality will only work well to predict
developmental dynamics (the 5%?), and to uncover evolutionary
mechanisms, but not to predict the combinatorics of those mechanisms.
We may never get beyond the indeterminacy in some aspects of both
quantum interactions and of empirical universal parameters and laws.
73. Adaptiveness Gap
Exponential Tech, Scaling Corps, and Tech-Using Adversaries vs.
Our Linear Psychology, Commands, Institutions, Laws, and Norms.
Our Vision is Linear. We Overexpect Early, and Underexpect Later.
We don’t Measure Network Effects, or Pre-secure Scaling Resources.
“10X Thinking” (Visioning 10X+ Change over Next 10 Yrs) can help.
74. Perennial Adaptation Curves:
First-Gen Dehumanizing, Third-Gen Empowering
New Technology Dehumanizes Us, At First
Kuznets Curve of Income Inequality
(Assuming Fixed Technology)
75. Good News:
Humanity is in the Age of Adaptive Peaks
Peak Autocracy was 1945
Peak Farmland (World) was 1970
Peak Deforestation (World) 1982
Peak Childbirth (World) was 2000
Peak CO2/Capita (US) was 2008
Peak Steel Use (World) was 2010
Peak Car Purchase (US) was 2015
Peak Oil Demand (World) is Now
World population will peak at 8.5-9 Billion in 2060. We are at 7.8B now.
Cities are far more sustainable and innovative per capita. World is 55% urban now. OECD: 85%
urban in 2100.
In 1981, 42% of the world lived in extreme poverty ($2/day). In 2016 it was 9%. Today it is roughly
6%.
Almost all natural resources are cheaper and environmentally cleaner today than they have ever
been.
Since 1995, world economic growth has had declining pollution/GDP and carbon emissions/GDP.
We still have many big problems of progress, including:
Climate change, collapsing wildlife populations, drought, conflicts, pandemics, rich-poor
divides…
But we’ll use accelerating science and tech, new networks, and new rulesets to solve them.
76. Wages and income reflect shared prosperity through 1978.
Since then, Exponentials have created a
growing rich poor income and asset gap).
Bad News:
Still Growing Power and Wealth Gaps
Rich/Poor, Capital/Labor, Invested/Uninvested,Urban/Rural,
High-Skill/Low Skill, Intangibles/Tangibles Corporations,
Advanced/Emerging Nations, many other Growing Divides.
Three K-Shaped Recoveries Since 1999
(Internet Bubble, GFC, Covid)
This pattern may continue.
77. Without Regulation and Redistribution:
Winner-Take-Most Network Dynamics
Devices: Cloud, Servers, PCs, Web, Smartphones, Embedded Systems
Companies: AWS, Google, Apple, Microsoft, IBM, Intel, Nvidia, Twilio
Example: Amazon Web Services
• Took 10 Years to Reach $10B
Annual Run Rate (22 yrs for Microsoft)
• Powers 1/3 of US Websites, 31% of
Cloud Infrastructure Market Share
• Can Build Many Critical Applications
with it in Minutes, Hours, Days.
• Runs 80% of all Containers (resource-
isolated code objects in virtual machines)
• 70+ Services in Analytics, Networking,
Mobile, Storage, Compute
• Has Decreased its Prices 75 Times
Since Launch
78. Tech Titans
(aka Superstars, Unicorns)
Even Tangibles leaders, like Amazon, are also heavily
intangible, with AWS and AI. Amazon spent more on
R&D in 2018 than US NIH ($36B).
Tech Titans are continually using Intangibles to take
over new sections of the Tangible economy.
Ex: Prime Air is now a FedEx and a UPS (Sep 2021).
Amz still smaller than Wal-Mart. How much longer?
Webb 2020
“Big Nine’s” (Goog, Amz, FB, Microsoft,
IBM, Apple, Alibaba, Baidu, Tencent)
core wealth is in their Platform Intangibles
(Software, Data, Patents, “Intelligence”)
79. Intelligence:
Grow and Use Open Source Platforms
Examples: GitHub, OpenAI, Amazon, Facebook, Alibaba, Uber, AirBnB,
ResearchGate, Wikipedia, ArXiv, Robinhood…
Example: GitHub and Open Source
GitHub has over 40M developers. 80% come from outside the US.
50M Free Repositories. Lots of DL Code (RNNs, GANs, etc.)
Why Did Google Put its AI Crown Jewels (TensorFlow) on GitHub in 2015?
It wanted to be at the center of the Open Development Community.
AI tools will commoditize and democratize, via open platforms.
80. “Empowering the Many” via:
Crowd funding, solving, collaborating, founding…
Intelligence:
Grow and Use Crowd-Benefiting Platforms
81. Intelligence:
Grow and Use Human-AI Teaming
China’s Ping An Good Doctor AI (2014) used 1K doctors to train AI on
text-based medical consults. System handles 75% of 700K consults/day.
The rest are routed to 5K specialist doctors, as on-demand contractors.
Riot Game’s Tribunal AI (2015) flags video game sexism, bigotry, and
harassment. Trained on 100M “votes”. High-rank players (not AI) set
punishments. Immediate 40% drop in abuse for 27M daily players.
Cornell’s Merlin AI (2013) began
with a curated dataset (eBird)
then added 800M crowdsourced
phone images, annotated w/ 3M
tags by ~5K amateurs. Unclear
classifiers are routed to ~500
Ornithology experts. Continually
gets better. Now identifies bird
sounds, too.
82. Intelligence:
Combat Bias with Learning + Diversity (Ensembles)
Diversity is as important as ability for:
• All poorly defined, “complex” problems (nonlinear optimization,
prediction, strategy, innovation, soft skills)
Human and Algorithmic Bias are best countered with:
• Diagnostics, Learning, and Cognitive/Model/Algorithmic Diversity
(“Ensemble AI”, IBM Watson, etc.). All Algos are Limited and
Biased. Ensembles are Less Limited, and “Multibiased.”
See also:
Superforecasting
Tetlock & Gardner 2016
Page 2008 Page 2018
Chap 3: “The Science of Many Models”
83. Know the Simulation Types:
1. Live (physical training exercises,
AR-enhanceable)
3. Constructive (dynamic maps,
battlespace displays, simulated
worlds and people, “Digital Twins”)
2. Virtual (People using VR training or
software models, ex: digital twins).
Each offers exponential capacity development opportunities
Intelligence:
Grow and Use Simulations, and Sim Training
84. This company has been
reported for not practicing
ethical fishing practices
This company has won
best fishing practices
5 years in a row!
Personal AIs (PAIs): A 2030’s Battleground
Late 2030’s: Personal AIs (PAIs)
2020’s: Conversational Agents
Google’s Assistant
Microsoft’s Cortana Amazon’s Alexa
Apple’s Siri
PAIs (Both Proprietary and Open Source) will
Increasingly Know Our Values and Tasks and Guide Us
In Viewing, Reading, Socializing, Buying, Voting, etc.
Very disruptive, with both opportunity and threat.
Her (film), 2013
Will Future AI “merge with our minds” via BCIs (Ex: Kernel, Neuralink)?
Perhaps in the late 22nd century. Today, steep tech and ethical barriers.
But PAIs will come soon, and are key to 21st century identity and power.
John Smart, Your Personal AI: A Five-Part Series, Medium, 2016.
85. What Are Your Networks Doing For You?
Val: $540B
Revs: $10B/Yr
Be an aggressive user of the best of your networks.
Often they aren’t the ones most people think are the best.
Val: $1B