Transcript Intro to Immortal Tech

SURVEY OF
ILE TECHNOLOGIES
V1.0
PHILOSOPHY
• functionalism of consciousness
• qualia has no direct physical
correlate
• a computer simulating the brain
will be conscious
SPACE-TIME
• a computer simulation executing just
1 instruction per 1000000 years will
experience the same qualitative
consciousness as the brain
qualia does not depend on
space- or time- scale
SUBSTRATE
• Jonah is fully
conscious inside
the whale
• his consciousness
does not contribute to
the whale’s consciousness
qualia does not depend on
substrate
LIFE IS
INFORMATION PATTERN
there is no clear-cut
distinction between
life and death
in post-upload world
substrate-hopping will be
the norm
MOORE’S LAW
“PROPHECY”
• in ~10-80 years computer = human
brain in power
• explosive increase in computational
intelligence will bring disruptive
change
• identity crisis for human race
SEXUAL REPRODUCTION
• sexual reproduction is not necessarily the
most competitive / desirable
• technological progress will continue,
driven by human desires
• balance between desirability and
competitiveness
NEED OF INTERFACING
seamless
AI / IA
integration
uploads
artificial intelligence &
intelligence augmentation
AI / IA
meat brains
2 APPROACHES TO
IMMORTALISM
1. uploading / neuroprosthesis
2. anti-ageing
cryonics
UPLOADING
• transfer memories and
personality to computer
• brain will be destroyed during
scanning
• philosophically sound but
somewhat unnatural
OBSTICLES
1. insufficient computational power
( computational power of brain = ? )
2. incomplete understanding of
neurobiology
3. uploads will not be accepted without high
confidence of isomorphism with biology
4. funding is a problem (uploading cannot
be done partially; lack of an incremental
strategy)
BRAIN EMULATION
requires:
1. electrophysiology (spiking models)
2. neural plasticity (modeling @
molecular level)
3. brain connectivity map (scanning)
4. whole-brain neurochemical
database (automated bioassay)
1. ELECTROPHYSIOLOGY
• electrophysiology is relatively wellunderstood
• realistic models tend to have too
many parameters
• dimensionality reduction: apply
statistical analysis to highly realistic
computer models to automatically
reduce # of parameters
2. NEUROBIOLOGY
•
•
•
•
simplify molecular models
memory encoded @ molecular level?
proteomics required
neural plasticity (especially synaptic)
is not fully understood
• complete elucidation of 1 pathway to
give estimate of total complexity
3. SCANNING
4. NEUROCHEMICAL
DATABASE
PRESERVING MEMORY
• scanning, vitrification, and
reconstruction may introduce
artifacts
• still, scanning is the best for
preserving information vs other
methods
• forgetting can be adaptive –
selective forgetting is preferred
“SOFT” UPLOADING
gradual replacement of the brain with
bionic components
CONFIGURATION
ALGORITHM
• Perceptron learning:
progressively adjusts
line to divide plane
according to sign of error
1
1
0
1
0
1
1
0
0
1
0
• extends to hyperplane and hypercube in
higher dimensions (n = # synapses)
• generalizes to gradient descent on highdimensional landscape (cost function) for
non-linear models
ANTI-AGING
• body is relatively unimportant
• within ~10-50 yrs all parts of the
body can be replaced by
synthetic / biological technologies
• heart, liver, blood, immune
system, etc… except the brain
BRAIN AGING
• myriad cellular processes
• pathways mostly unknown
• replacement is more feasible
than repairing
• progressive deviation from
optimality
BRAIN REJUVENATION
requirements:
1. integration with host
2. suppression of rejection
3. innervation of targets –
requires re-expression of
developmental cues
4. proper neurochemical
function
AVAILABLE TECHNIQUES
1.
2.
3.
4.
5.
6.
xenotransplantation
neural stem cells
regeneration
neurogenesis
genetic approaches
nanomedicine
1. XENOTRANSPLANTATION
recent progress:
 immunosuppression of host
 immunomodification of graft
 integration with host
 synapse formation
2. NEURAL STEM CELLS
recent progress:
 long-range innervation from
implant to host
 cryopreservation
• developmental niche is lost in
adult brain
3. REGENERATION
1. facilitate axonal regeneration
using Schwann cells and
olfactory enshealthing cells
2. applicable to spinal cord
injuries and peripherals
4. ADULT NEUROGENESIS
• up-regulate to replenish
neurons in the brain
in hippocampus, and
sub-ventricular
zone (SVZ)
migrate to
cortex
4. GENETIC APPROACHES
• use of retroviral vectors to
modify genetic code
• modification of the cells’
neurochemical properties
such as axonal guidance
5. NANOMEDICINE
• requires sophisticated
nanotechnology which is
uncertain
• may become unnecessary
because of uploading
CONCLUSION
high-complexity road blocks in both
approaches:
• uploading: neural plasticity &
information-encoding at molecular
level is murky
• brain rejuvenation: re-creation of
developmental “niche”
success
linear approach is bound to fail
complexity barrier
funding
must concentrate resources to
systematically solve problem
© Dec/2003