2. What Is It? How can it occur?
The notion of some radical technological shift beyond
which the world is unimaginable
Visions of the future
Just the continuation of more and better gadgets and
faster computers
Accelerating Returns
The rate of change, of innovation, is accelerating, even the rate
of its acceleration is accelerating.
Event Horizon
Intelligence Explosion
The advent of greater than human intelligence
Apacalyptism
Geek religion?
3. Models of Singularity
Advent of greater than human intelligence
AGI
Brain emulation
Augmenting human intelligence
Accelerating change
Which also makes brain emulation and workable
AI nearly inevitable
But some say just the change itself will grow
beyond our ability to understand or predict
4. Accelerating Returns
Examination of multiple events and capabilities
shows exponential increase
All are driven by (and contribute to) intelligence
Rate of inventions, innovation
Adoption rates of innovation
Increase in computational ability
Increase in communication ability and use
Time between paradigm shifts
Computational power per unit cost
5. Accelerating Returns
Core Claim
Technological change feeds on itself and
therefore accelerates. Our past or current rate of
change is not a good predictor of the future rate
of change.
Strong Claim
Technological change follows smooth curves,
typically exponential. Therefore we can predict
with some precision when various changes will
arrive and cross key thresholds, like the creation of
AI.
Advocates
Ray Kurzweil, Alvin Toffler, John Smart
14. Computational Increase
Predictions
Human brain capability (2 * 10^16 cps) for $1000
around 2023
Human brain capability (2 * 10^16 cps) for one
cent around 2037
Human race capability (2 * 10^28 cps) for $1000
around 2049
Human race capability (2 * 10^28 cps) for one
cent around 2059
15. Computational Acceleration
Blue Gene will have 5% of the computational
power needed to match a human brain
In the next five years or so supercomputers will
match the computing capacity of the human
brain
By about 2030 such a machine will cost around
$1.
What happens when you put 1000 of them to
work 24x7 on say the AGI problem or MNT or
ending aging?
16. Limits of Human Intelligence
20 – 500Hz. Neuron firing rate
Lately there is evidence than the pattern of firings is the important
information unit
Very few aspects of a problem may be held in conscious awareness at one
time
Very slow learning rates
Faulty memory
Limited communication ability with other minds
Squishy deteriorating mind implementation
Poor scalability
Poor transfer of knowledge
Very slow replication
17. Past Intelligence
Augmentation
Communication. Starting with speech
Writing – an early augmentation of human intelligence
Stable cultures respectful of knowledge and encouraging to
innovation
Dissemination of learning
Science and Mathematics
Reality respecting culture
Computation
Gathering and doing more with information ever faster
Device Convergence
18. Future Intelligence Augmentation
Ubiquitous computing
Exo-cortex
Wearable or embedded computing /
communication
Brain – computer interface
Upgrading the brain with technology
Bio-chemical upgrading
Upload
19. Event Horizon
Core claim
Soon we will not be the greatest intelligence on the
planet
Then the changes in our world will no longer be in our
control or understadable and predictable by us
Strong claim
To understand what a superintelligence would do you
would have to be that intelligent. Thus the future after
that is totally unpredictable.
Advocates
Vernon Vinge
20. Greater Than Human
Intelligence
"Let an ultraintelligent machine be defined as a
machine that can far surpass all the intellectual
activities of any man however clever. Since the
design of machines is one of these intellectual
activities, an ultraintelligent machine could
design even better machines; there would then
unquestionably be an 'intelligence explosion,'
and the intelligence of man would be left far
behind. Thus the first ultraintelligent machine is
the last invention that man need ever make.”
- I. J. Good 1965
21. Intelligence Explosion
Core Claim
Creating minds significantly smarter than humans
closes the loop creating a positive feeback cycle.
Strong Claim
The positive feedback cycle goes FOOM, each
improvement triggering the next.
Superintelligence up to the limits of the laws of
physics develops rapidly.
Advocates:
I.J. Good, Eliezer Yudkowsky
22. What does Singularity Mean for Us?
Short Range
Employment / Work
Cures for all disease including aging
Near limitless abundance
Economics
Political
Backlash
Sociological
23. Impact of Human Equivalent
Machine Intelligence
Assume
Human intelligence without biological limitations and
distractions
Can focus and work 24 x 7
Ultimately very cheap to own and easy to replicate any unit’s
knowledge
Results
Almost all human labor, especially intellectual becomes
economically superfluous
Progress increases substantially when entire armies can be
deployed on projects
So..
economics must change a great deal for all humans to
continue to partake in results
Our view of ourselves must change
Maybe we join the AIs and become uploads
24. > Human Intelligence Impact
Much that goes on is totally opaque to even
augmented humans
Humans even transhumans are not in charge,
are second class citizens
Are they respected? Kept around?
Our history with other species is not encouraging
Are humans irrelevant to AGIs?
Is there an objective ethical argument for treating
lesser intelligences well?
Is going extinct to usher in a greater intelligence
“OK”?
25. What does Singularity Mean for us
Long term
Do we survive?
If we do not are we happy to have produced
wonderful “mind children” as Moravec suggests?
Are we recognizable as human?
Does it matter?
What of those who do not wish to change?
Unlimited future..
Unchecked or unwise recursively improving AI goos
the universe
26. What Could Stop Singularity?
Turns out to be unnoticeable and no big deal
Harsh Condition Slow, Halt, Reverse Progress
Unexpected limits and difficulties
27. Harsh Things Happen
Economic Disaster
Some believe this will take a decade or two to
clean up
Energy Disaster
Easily a decade to largely replace oil
Existential Risk
Major War
28. What would Utopia take?
We become super bright, compassionate and
enlightened
AGI is Friendly and bring us the best we can
possibly have as quickly as it deems best for us to
have it
29. AGI ends up Friendly
because
Its own decision based on:
We are it progenitors
It finds us curious or amusing
Its ethics lead it to treat lesser intelligences well
Some really bright human hackers gave it
unbreakable limts or top level utility function
30. Is the best we can get
utopia?
What is utopia?
End of all suffering?
Does this mean end of wanting what you don’t and
perhaps can’t yet have?
Does it mean the end of negative feedback or
consequences no matter what you do?
Do you instantly become enlightened?
End of material lack
Beyond lack to people stop wanting?
Is utopia nothing left to do?
Can humans be happy as second class or so
totally outclassed as to be beyond
comprehension?
Notas do Editor
Cells (10**9), body parts (10 ** 7.8), mamals (10**7), primates (10** 6.7), humanoids (10 ** 6), Homo sapiens (10 ** 5) , Stone Tool (10 ** 4), Iron (10 ** 3), 10 ** 2.8, printing 10 ** 2.5, TV 10 ** 1.8, computers 10 ** 1.51, 10 ** 1
Doubling time for use. Electricity 1870 – 45y, Telephone, 1890 5 – 35 y, radio 1898 28 y, Television 1930 25 y, PC 1979 18 y, mobile phone 1984 13 y, Internet 1991 3 y