Kurzweil - The Singularity Is Near
Ch 1 - The Six Epochs
- The singularity = point at which technological change is so rapid, human life transforms irreversibly
- Neither utopian nor dystopian
- Based on the idea of exponential growth
- In several decades, artificial intelligence will be as intelligent and proficient as humans
A. The Intuitive Linear View Vs. the Historical Exponential View
- Human progress is exponential, not linear - i.e. doubles rather than adds a static amount
- The tendency is to overestimate progress in the short term and underestimate in the long term
B. The Six Epochs
- Evolution through indirection - "Each stage or epoch uses the information-processing methods of the previous epoch to create the next" (14)
- One: Physics and Chemistry
- information in atomic structures
- basic laws of physics
- think pure, physical world
- Two: Biology and DNA
- biological systems form out of physical molecules
- Three: Brains
- detection of information begins
- formed from DNA, which is formed from molecules
- Four: Technology
- human-created
- ultimately capable of sensing/storing/evaluating information (think webcams or text-recognition software)
- Five: Merger of Technology and Human Intelligence
- the limits of biology are transcended
- Six: The Universe Wakes Up
- matter becomes saturated with intelligence - not just brains, but molecules themselves
C. The Singularity is Near
- "Intelligence explosion" - what Kurzweil describes as technology creating its own next generation (explored in next chapters)
- Technology can reproduce infinitely as soon as we transcend the limits of human biology
- The singularity is coming, soon - a matter of decades
Ch 2 - A Theory of Technology Evolution
A. The Law of Accelerating Returns
- Law of accelerating returns = exponential growth of progress
- The nature of order
- complexity does not necessarily mean complicated
- complexity = the minimum amount of information needed to represent a system - meaningful, non-random, and predictable
- complexity does not constitute progress on its own, it needs to be accompanied by order
- order = information that fits a purpose
- Life cycle of a paradigm
- 1. Slow growth
- 2. Rapid, explosive growth
- 3. Leveling off
- For a paradigm to evolve, it needs some amount of chaos and the predecessors to build off of
- At its core, evolution increases observational distance and duration - we are able to observe more information for longer periods of time
B. Life cycle of technology
- The Cycle:
- 1. Precursor stage - prerequisites exist, but inventing hasn't happened
- 2. Invention
- 3. Development
- 4. Maturity
- 5. Upstart threatens technology but doesn't succeed
- 6. New threat succeeds in overtaking old technology
- 7. Technology yields to antiquity
C. Moore's Law and Beyond
- Moore's Law = we will double the amount of transistors on a circuit every two years
- Is this a self-fulfilling prophecy? (e.g. we set this as a goal and work towards it)
- "There is a gentle, but unmistakable exponential growth in the rate of exponential growth" (pg. 69)
- According to Moore's Law expanded, supercomputers will match human brains by 2020
D. DNA Sequencing, Memory, Communication, the Internet, and Miniaturization
- GNR - genetics, nanotechnology, robotics; a new era in technology that we are rapidly entering
- Results from intertwining technologies, not just computers
- Communications technology is exponentially growing as well - sets the stage for supercomputers
- S-curves:
- Exponential growth followed by an evening out
- Moore's Law can be expanded to think about miniaturization - the smaller the better
- Informational viewpoint - the universe functions essentially as software (reading and interpreting information), not particles and forces
- Example: Wolfram's cellular automata - "Rule 110"
- Cellular automata that produces complex, unpredictable models from simple, deterministic rules
- Rule 110
- Take-away point: simple systems can produce complex results
E. The Singularity as Economic Imperative
- Economics are evolving - both result of and fuel for technological evolution
Ch. 3 - Achieving the Computational Capacity of the Human Brain
A. The 6th Paradigm of Computing Technology
- Computing must become 3-dimensional to reach the singularity
- We're limited by 2D computing
- This means nano-technology needs to happen
- Key to nanotechnology: self-assembly and self-organization - technology that emulates DNA
B. The Computational Capacity of the Human Brain
- The human brain processes 1014 instructions per second
- Another way to look at it is 1014-1015 CPS
- To simulate human personality, we would need to model individual neurons, each of which has 103 intraneural connections
C. The Limits of Computation
- Exponential trends tend to eventually reach a limit
- Brains run as parallel, low power computers, whereas modern computers run one thing at a time (not parallel) at high speeds
- Example: How Smart is a Rock?
- The information stored in the atoms of a rock represent 1027 bits of memory, but it's unused because the organization is largely random
- Irreversible processes (how computers usually run) mean that information is deleted - the deletion requires energy and releases heat. Reversible processes simply move data to another location - saves energy and results in faster processing and less heat created
- Ultimately, if we start building parallel, 3D computers, the limits of computation are nil
Ch. 4 - Achieving the Software of Human Intelligence
A. Reverse Engineering the Brain: An Overview of the Task
- Scientists create computers in ways that do not mirror the human brain - we lack tools to do otherwise
- The first step is imaging the brain effectively - moving beyond MRI's to real-time, high resolution, 3D imaging
- "A principal assumption underlying the expectation of the Singularity is that nonbiological mediums will be able to emulate the richness, subtlety, and depth of human thinking" (pg. 145)
- Computers have a few legs up, however - primarily ease of sharing info
- Technology is quickly moving towards emulating biology
- The human brain measured in computational terms contains about as much information as Microsoft Word
- Take-away point again: complex results come out of simple systems
B. In the Human Brain Different From a Computer?
- Human brains work in parallel, computers work non-parallel. Differences include:
- Computers can recognize patterns that human brains can't
- Speed
- Brain combines digital and analog
- Brain rewires itself, adapts
- Most details in the brain are random
- Brain uses 'emergent' properties
- Brain is created imperfectly
- Brains contradict themselves
- Brains evolve
- Brains have a very specific architecture (i.e. certain areas perform certain tasks)
- Brain structures are simpler at higher levels (e.g. lobes versus neuron structure)
C. Peering Into the Brain
- Brain imaging techniques like PET and fMRI are subtractive (comparing the active brain to the non-active brain)
- Brain imaging improvements requires higher resolution
- Nanobots will allow us to scan from the inside
- Nano-brain imaging technology will be available by 2020
D. Building Models of the Brain
- Lower-level is often more complex (think cells vs. brain architecture)
- The ultimate question - are our thinking processes capable of understanding themselves?
- Subcellular components and the synapse are the primary units of the brain
- "I think therefore I am" (173) and brain plasticity
- Labs have successfully simulated input/output regions of the brain (e.g. robots that can see or hear)
E. Interfacing the Brain and Machines
- Robots can be controlled by thought alone in some experiments - prosthetics and cochlear implants are everyday examples of this kind of technology
F. Uploading the Human Brain
- We center around physical thoughts and desires, and as such uploading a person requires a body
- What are the limitations of the Turing Test? Would an uploaded person really be 'you'?
Ch. 5 - GNR
- Genetics, Nanotechnology, Robotics
- Biology will never match technological progress
A. Genetics: Intersection of Information and Biology
- Aging is a "problem to be overcome" (210) - what can we say about this?
- Disease as a war - who has access? pg 212
- Moving into an age of designer babies and designer parents
- Thinking of cloning and eugenics - what are the implications of the genetics revolution? (see pg. 221)
B. Nanotechnology: Intersection of Info and Physical World
- Nanotechnology will allow us to rebuild our bodies
- Molecular assemblers - possible? (think Star Trek)
- These will solve the problem of self-replication
- We run into the "fat fingers" problem - construction tools that are too large to create tiny tech
- Nanotech manufacturing will potentially reduce emissions - "minimal consequences" for the environment
- Nanobots in the bloodstream = the ultimate medical intervention - highly specific illness treatment
C. Robotics: Strong AI
- AI (artificial intelligence) that reaches and exceeds human intelligence is the goal
- Ways AI can exceed human intelligence:
- Knowledge sharing
- Pattern recognition
- Resource pooling
- Exacting memory
- Consistent performance
- What about runaway AI?
- No need to fear - humans can't enhance their own intelligence significantly, neither will human-esque robots
- Eventually it will be possible, but not right away
- "Narrow AI" already exists - AI that performs one specific task (like playing chess or observing surroundings)
- Turing Test - we will pass the threshold long before we realize it
Ch. 6 - The Impact
A. On the Human Body
- qt. 301
- Eating for pleasure w/ technology that allows efficient digestion, limits intake of bad things
- "Programmable blood"
- Allows us to go hours without respiration
- Better hemostasis
- Better antibiotics
- Endless possibilities to extend life
B. On the Human Brain
- We'll achieve immersive virtual reality by 2010 - embedded in our clothes and accessories
- By 2030 - nanobots exist in our bodies and allow for instant access to virtual reality - and instant access back out
- Identity once tied to our bodies will cease to exist - what happens when our bodies are changeable?
C. On Human Longevity
- Eliminating medically preventable diseases will extend life expectancy 500+ years
- Conceptualizing of software (brain) transfers to new hardware (body)
- Counterpoint - is death desirable? What role does death play in giving meaning to life?
- Problems with archiving
*Archiving technology quickly becomes outdated (think floppy discs, now essentially unusable)
*"Information lasts only as long as someone cares about it" (329)
D. On Warfare: The Remote, Robotic, Robust, Size-reduced, VR
- Casualties of war will become increasingly less
- Better armor, better weapons
- Contrast Kurzweil with transhumanism - why "better" war when we could implement technology to prevent war?
E. On Learning
- Decentralized education - i.e. everyone has access to the same information
- This means virtual, web-based learning (already progressing towards this via MOOCs)
F. On Work
- qt. 338
- We'll have an increasing need for protection of intellectual property as information becomes more equated with value
G. On Play
- Play = the mother of work
- Play will evolve to unthought of heights
H. On The Intelligent Destiny of the Cosmos
- ETI = extra terrestrial intelligence
- If ETIs exist, we should have noticed them by now (or rather, they should have noticed us). However, the likelihood that they exist is extremely high
- This is the Fermi Paradox
- Any intelligence we do find will likely be mechanized
- Enter the idea of the mutliverse
- In short, eventually everything in the universe will become part of a singular mind
Ch. 7 - Ich Bin Ein Singularitarian
- The singularity is not a replacement of faith
- What does it mean to a Singularitarian?
- We have the means to live forever
- There is something inherently wrong with our current biological bodies qt. 371
- We need a body, but soon we will be able to change our body at will
- Knowledge is precious, but information is not knowledge
- Death is an avoidable tragedy
- The purpose of life is to move towards order
*How do we decide the distinction between human and posthuman?
A. The Vexing Question of Consciousness
- Will machines be capable of emotional/spiritual experience?
- They will claim to, but how do their experiences relate to those of humans?
- qt. 378
- How can we experience the subjective reality of other humans?
- The question of consciousness cannot be answered empirically for humans or robots
B. Who am I? What am I?
- How much of identity is tied to physical embodiment?
- Technically speaking, our cells are completely replaced every month
- "a pattern of matter and energy that persists over time" - is this the definition of personhood?
- What about duplicates? What does it mean to be a person if another copy exists?
C. The Singularity as Transcendence
- Transcendence is a part of spirituality
- Transcendence is everywhere
- Going beyond ordinary powers of the material world via patterns is Singularitarian transcendence
- block qt. 389
Ch. 8 - The Deeply Intertwined Promise and Peril of GNR
- With access to information comes the danger of how that information is used
- We might consider limiting development of perilous technologies to combat this danger - but how do we know which technologies could be harmful?
A. Intertwined Benefits
- "Technology empowers both our creative and destructive natures"
- qt. 396
- We'll encounter the ability to live our desired lives
- Elimination of poverty, illness, and more
B. And Dangers
- Technology is already often used for destructive means
- Biological weapons, for instance
- With more advanced weaponry, however, comes more advanced defense/reaction to threats
- "Grey-goo scenario - in which nanobots reproduce uncontrollable at the cost of biological life - unlikely to occur with proper precautions
C. A Panoply of Existential Risks
- Bostram's Categorization of Risks - existential risks are very bad, threaten all of human existence
- The only existential risk we've faced is the threat of a nuclear winter, biological weapons
- GNR will end biological existential threats but bring in new ones
- Precautionary principle vs. proactionary principle - balancing the risks of action and inaction
- We are unlikely to stumble onto an existential risk in our progress
- On the other hand, progress is inevitable
D. Preparing the Defenses
- Strong AI can be guided to be friendly
- We must move forward to solve current problems, even if there is risk involved
E. The Idea of Relinquishment
- How much technology should we relinquish?
- Certain fields could be abandoned (for instance, nanotechnology)
- Or, certain ideas (e.g. self-replicating nanotechnology)
- Update/upgrade regulations make sense - they're necessary to safety
- Fundamentalism poses a barrier - of all kinds, not just religious
F. Development of Defensive Technology and the Impact of Regulation
- Again, with growing threats come growing defenses