Robert L. Blum, MD, PhD

 

Home

Photos

Sphere of Interest

WebBrain: AI
neurosci psych

Stanford Brain Lecture Notes

The RX Project:
Robotic Discovery

CV Biblio (1985)

Contact


Index of Essays

Psychology &
Neuroscience brain-icon

Computer Science,
Robotics, and AI
pr2-icon

Health & Biotech
insulin-icon

Earth Wisdom: Biosphere and Universe earth-icon

Be Saved by Bob!!!
(And Other Balms )
blowfish


Optimal Nutrition:
Are Fats Killers
or Saviors?

The Mystery of CONSCIOUSNESS

Consciousness Video:
Who, What, When?

Near Death Experiences: In the Desert With Pim Van Lommel

Is the UNIVERSE
Fine-Tuned for Life?

EUV 2013 - Future of Moore's Law

BAM: Brain Activity Map of Spikes

Beating Jeopardy!
What is Watson?
AI Overlord or Tool?

SETI: Search for Extraterrestrial Intelligence

KEPLER Seeks Earth-like Worlds

STEVE PINKER in the Amazon: photos

Billion Year Plan:
AI Formulation

AI Awakens

CONSCIOUSNESS as Global Resonance

SEAN's Accident

Coronary Artery CT
Scan: A Life Saver

Book Review: TRANSCEND

Book Review:
Create a Mind

Does Drug X
REALLY WORK?

TRANSCEND
DRUGS!!!

Total Recall:
Everything, Always

Ralph Triumphs:
Elbot Cheers

Scientists &
Evangelicals Unite

Thomas Berry,
Geologian: Obituary

Calorie Restriction
Works in Monkeys!

TheBrain &
WebBrain: Review

 

Singularity Summit 2010

Aug 14,15: San Francisco

 Notes and Transcripts of Lectures

 

RAY-Singularity

 

            I just attended the 2010 Singularity Summit  (SS10), which took place

Saturday and Sunday,  August  14th and 15th at the Hyatt Regency in San Francisco.

This conference is a favorite:  this has been my fourth.

 

The leading luminary is futurist Ray Kurzweil, whose 2005 book

The Singularity is Near was the best exposition to date of the nature,

evidence for, and seeming inevitability of a technological singularity:

 machines taking over as the key innovators on the planet, for better or for worse.

 (This might happen when machines can autonomously design and build

 their successors: a closed loop of ever-accelerating intelligence augmentation.)

 

            I have about 48 hours before I jump off the grid completely

for two weeks  and lead life as John Muir did in the Sierras

100 years ago (except for  the down sleeping bag and freeze-dried food).

Like the Savage in Aldous Huxley's  Brave New World, I will bask in the beauty

of a natural world, untouched by humans.But unlike the Savage,

 I long to watch the future unfold, perhaps with untold delights:

an extension of human lifespan, freedom from diseases and other bodily constraints,

undreamed marvels of technology, and perhaps unparalleled cooperation and

 wisdom among nations.  There are, of course, perils aplenty.

 BACKPACK

            Since I will soon depart, I have only had time to post

 my detailed lecture notes, seen below. I have analog tapes of these talks

but haven’t done full transcriptions yet. In six months or so videos will be available.

 In the meantime, perhaps these will tide you over.  They appear in chrono order.

 Use these notes in conjunction with the presenters’ abstracts,

which appear here on the web site of  SS10.  Feel free to disseminate.

 

JUMPS to my notes below are here:

 

Saturday August 14th 2010


Ray Kurzweil

 

Sunday, August 15th 2010

 

(I was unable to attend the talks by Srinivasan, Heber-Katz, Goel, Pepperberg, and Randi)

 

 

SEAN MCCABE

      Sean did introductions for the speakers.  These were great infotainment.

 Sean reminds me of the character "Guy" in Galaxy Quest, who introduced

 the starship crew at a fan conference.  Sean, I recommend the "echo style" to you:

"RAY RAY Ray Ray ray ray, KURZWEIL KURZWEIL Kurzweil Kurzweil kurzweil kurzweil."

 

 

********************

Michael Vassar

 

MICHAEL VASSAR (Singularity Institute) Saturday 9:05 AM

 

The DARWINIAN METHOD

 

(Michael is President of the Singularity Institute.)

 

This talk is on the history of science and contrasts

the earlier,  armchair "Scholarly Method" of science  with the

evidence-based Enlightenment  Method.

Science sometimes makes fruitful use of a synthesis of these.

 

What is science?

You can't just make Gothic cathedrals without at least some rationality.

So, even if it wasn't science, there was considerable rationality back then.

 

Knossos had printing.

Indus Valley had running water

 

pre- 17th century Europe in comparison was a backwater,

but did have really good art.

 

Archimedes clearly had  rational innovations in Classical times.

using the displacement of water as a test for the purity of gold.

Michael, however claims Archimedes was not doing science.

Why wasn't this science?  no hypothesis testing,  no organized literature and publications.

 

Etruscans thought  that lightning was caused by clouds banging

into one another (rather than the gods); so, at least that shows some progress.

 

"Scholarly Scientific Method"

Read broadly, it consists of the following: 

write abt diverse  life experiences,  and look for convergence in what people say.

That was the main method in ancient Greece and the Middle Ages.

 

How to produce great child scientists?

Start with kids who are unusually neotoneous

(this is maintaining child-like curiosity and openness into later years)

Then  keep  error rates high by exposing them to new material and

new experiences all the time;

Nice slide of cute kid with glasses. 

how to create scholars?

 get kids that notice subtle patterns and give them diverse experiences;

  

Why scholarship fails...  one reason it fails is because of correlated errors.

wonderful cartoon :   Intelligent Design for Dummies

 "a reference for the rest of us"... by Dan Dog.

You need separate communities to notice different things and to

 eliminate  spurious correlations.

  

There is a reporting bias towards surprising claims.

This is a fundamental bias :  surprises get published. 

If you report a result saying "Sky is blue"  no one will print it.

 

What changed in 17th century...  radical skepticism..

in ancient Greece it simply meant  considering both sides

 

Enlightenment Science: 

build long chains of reasoning

 from seemingly solid foundations  and look for surprising conclusions.

Rousseau ... savages are better off than  Europeans 

(apparently Ben Franklin also shared this belief)
Michael V emphasizes that this was espec true in the 1700's, evidenced by net migration.

 

Steve Pinker in his TED Talk clarifies that at present

more than half of males in many primitive tribes  DIE  from  homicide.

 If this was true in larger society,

we would have had  billions of 20th Century deaths instead of only  millions.

 

Darwin and Wallace...  working independently in the 19th Century

arrive at same conclusions; independent confirmation is key to science.

 

****************

 

 

MetamanGreg Stock

 

GREG STOCK  (CEO of Signum Bioscience, biophysicist and best-selling author:

of the Book of Questions (sold 2.5M copies) and Metaman (a favorite of mine)

 

EVOLUTION and The POSTHUMAN FUTURE  (Saturday:9:45 AM)

 

(a stunningly good talk: many themes reminiscent of his book Metaman)

 

He cites "Marooned  in Real Time" by Vernor Vinge

 (a singularity pioneer, math prof, and sci fi doyen)

an empty landscape devoid of humans.  (one poss of the future)

 

Greg asks "so, where are the wonder drugs?"

Problem is that it takes years to do each clinical trial;

  The FDA needs to address this problem.

He (and the rest of us) want to extend our life spans, but will he (in his sixties)

 live  long enough to be able to take advantage of the "next bridge" 

 that Kurzweil cites (drugs that implement current genomic discoveries)

 

Is it likely that the Singularity will lead to love and to a  triumph of human values?

 It seems more probably that it will lead to human extinction (in some form). 

It seems quite doubtful that we will be able to reach thru the singularity in order

to preserve human values/love etc.

We are meat,flesh and blood.

He reads a v poignant quote from Wm Butler Yeats 1926

"Consume my heart away; sick with desire and fastened to a dying animal..."

 

His company Signum has been targeting Alzheimers;

idea is to preserve something (a functioning brain)

that is worth freezing (cryo preservation).

If u live to age 85... 50% chance of Alzheimers histologic changes

 

At Signum we look at a key molecule PP2A

It removes phosphates from regular proteins;

It has broad effects on master regulatory proteins.

One of their great findings is that coffee activates PP2A.

fun slide showing fifties housewife drinking coffee:

 COFFEE -  you can sleep when you're dead!

 

Has some evidence that coffee consumption decreases Alzheimers

by 50% and  also adult diabetes,.

The effect is apparently NOT related to the caffeine content

 (don't get the effect from just the caffeine).

Their lead compound is Sig 1012 - a coffee bean extract.

(a serotonin like molecule  attached to a lipid).

In a Parkinsonian mouse model, you see improved nesting behavior

(tearing and collecting bits of paper to create a nest).

Should soon be able to move on to human trials, since coffee bean extract

is  GRAS  (generally regarded as safe by FDA)... coffee bean is a food.

They will test this to prevent Alzheimers.

 

Will we ourselves be transcended? eg by some combo of  carbon + silicon.

 

Evolution is not static; its methods,are also constantly evolving;

he shows slides of earthworms doing it;  Japanese  beetles and zebras,

 squids, and  hover fliesin amorous embrace.

 

Social evolution is a kind of evolution in which ideas are competing (memes)

Essentials of evolution: reproduction, variation, differential survival.

With humans (unlike Moore's Law) there are no exponentials in improvement;

the generation time is just too damn long.

 

Transitions and Jumps in organization complexity: blue green algae first...

Warrawona  group ...  3.5 B yrs ago.... then on to E Coli  then

on to Eukaryotes 2.2 B years ago.

 

Then abt 1.2 B years ago-  multicellularity.

Jelly Fish... then ….

 

Now aplanet-wide superorganism.:  (Metaman… human civilization, writ large)

Great slides of  internet pathways,

Earth at night,

Hong Kong.... great picture...

that substantiate the increasing size and complexity of our current superorganism.

 

Comparison to a termite mound with the queen termite

This is a new global union, a planetary superorganism....  it is beyond metaphor

(This was Greg's theme in Metaman.)

Metaman has tightly orchestrate internal networks; digestive and circ systems,

a cns,  it is robust and redundant, it makes use of brutal internal competition;

and  benefits from breakthrus in materials...

 

 

 

 

Example of a breakthru  560 M yrs ago in materials.

New material used in bio orgs...  CALCIUM PHOSPHATE

used to create bones,  skeletal structure; 

essential for the transition to METAZOANS.

 

And now a new transition that combines complex bio + complex nonbio (materials)

 

What is the Nature of transitions..?

new materials SUPPLEMENT the old ones

new levels subsume old ones...

get localized homeostatic environments:  eg bee hives,  termite nests,

and human houses and cities.

As an example, when he travels from his house in Princeton, NJ, to Korea,

a roof is constantly over his head (we have built protected environments).

 

Observe the power of evolution over just a few thousands of years...

A Grey Wolf....  was evolved into pet dogs...

a huge diversity of dogs:  chihuahuas, blood hounds et al

 

Humans may undergo a similar transformation.

 

Free Range humans may only remain in tiny little back eddies;

he says we will NOT be free range anymore

 

In Evolution our components are reused and refashioned.

 

Although a single neuron in the amygdala network may be part of the

network that mediates fear/ emotion;  it itself knows nothing about fear.

May be the same as us in relation to the entirety of human civilization/ metaman.

 

 

Look at subprime crisis...  we are terrible at predicting future;

the future may get v weird, v quickly.

 

Does it matter when the Singularity happens?  Not really.

It does not matter whether it’s in 20, 50 or 1000 years.

It's an instant in the sweep of time.

(Of course, he and I would like to see it in our lifetimes.)

 

Human ethics and values are not unique to us.

Our species has kinship groups and extended social ties but that is not unique.

Chimps have it all... sharing, empathy.

slide: Mourning Dorothy's death... chimps watching the

dead body/ funeral of one of their respected older members.

chimp consoling a crying child  (or perhaps the chimp's presence is causing the child to cry.)

 

It's a mistake to think we are paragons of rationality.

  We are hugely driven by raw emotion.

Consider that Lebanon  was at peace for 2000 yrs then war and destruction;

same as in Bosnia and Zrebreniza.

"Marx was right about socialism... he just had the wrong species"   E O Wilson quote,

said abt his favorite species ... the ants.

 

What might  values be in a post-human, cybernetic age:

competition will be extremely strong;  sex will be vestigial,

copies will be  cheap;  death is meaningless.

Uploading (if poss) will disengage us from  our bodies.

 

A slide of an uploading human ...  he comments...

v evocative of Christian symbolism and images... Christ-like transcendence;

what would such things have in common with humans... v little...

 

What will we become?

Nick Bostrom.quote:  there will be a gradual elimination

 of all forms of beings that we care about.

 

Carl Shuman...   quite inhuman values...

Aren't we kidding ourselves to think that we can possibly intervene

 in this process and stop evolution.

 

Nick Bostrom: a super entity might stop the presses  to avoid competition...

the singleton rule... to protect against its own demise by competitors.

the singleton controls the space.

We derive our specialness from our environment  (materials, culture, knowledge)

 

People currently are concerned about clones: 

some are exuberant about them, others have angst

But compared to the Singularity,

the whole issue is pathetically unimportant and quaint.

 

Consider issues of slippery slopes:,  losses of identities,

abhorrent values and changes,  perverted values..

We should possibly be worried about it; 

He shows the Edvard Munch painting: Angst.

 

Preserving values as we we pass THRU the singularity...

has similar character  (angst vs exuberance)

some super AI may manage and control it....

this is perhaps the wish of the Singularity Institute;.

he thinks it’s impossible.

a complex, emergent realm careening toward the future

 will  go its own way, regardless of what we do or wish.

 

Final quote:  evolution will come from the nimble and the bold.

 

a 430  BC  Thucydides quote ...  the bravest will shape the future.

 

*****************

Ray Kurzweil

 

RAY KURZWEIL (Inventor and Leading Luminary of Singularity)

 

            (Ray is on vacation at home in Massachusetts;

  he  uses  video link to give his lecture;

quality of video is excellent, and he can see us as well.)

 

 The MIND and How to Build One  (Saturday: 11AM)

 

Ray begins by addressing one of his critics -

John Horgan,  former Editor  in Chief of Sci. American.

John claims that the brain is too complicated for humans to figure out and to

emphasize his point he shows a highly complex brain with millions of neurons.

Ray says he located that image and it's actually not a real brain but

a computer-generated brain (eg made by Blue Brain)  LOL.

 

Another early critic was Prof. Tommy Poggio, who is head of machine vision at MIT.

Around 2000 or so,Tommy told Ray - you know, we're not getting any info

from brain studies that can help us design machine vision

Ray predicted that it would happen.  When Ray saw Tommy in 2006

at the 50 year anniversary celebration of the Dartmouth AI Conference (Minsky/ McCarthy),

Tommy said ... you were right.  Poggio had built a machine vision system based on

the HMAX algorithm which is inspired by the wiring of complex cells in visual V1

to simple cells in V1... it outperforms humans in certain visual tasks (with very brief duration

exposure).

 

            Can we design artifacts before we fully understand the principles of the brain?

Yes.  Example is Bernoulli principle ,,,  entire world of aviation is based on it,

 but still not fully understood.

Once we understand algorithms of cns we can add them to our AI toolkit.

 

            We seem to have a stable estimate of the compute power of the brain:

10**14 to 10**16 calcs per second.  IBM will have  a supercomputer that cranks out

10**16  calc/sec within a few months- and  will be shipping.

 

            Justin Rattner  (CTO of Intel)  says that INTEL

 has 3d  electrical circuits working in the lab.

possible  successor to planar  CMOS.  Then, of course, 

cloud computing is almost here ... use vast array of machines sitting idle.

 

Ray's doubters: 

 Doug Hoffstedter,

 Jaron Lanier;  Jaron (at Cal Berkeley) says current machines

 are not doing real AI and are not conscious -

but that is really a straw man.  Nobody claims that current machines

 are consc or doing human level AI.

See Jaron's op-ed in the New York Times.

 

Info Tech  (IT)  is exponential; so,   30 doubling means an expansion by a billion... 

amazingly, no discontinuities in Moore's Law despite world wars, depressions, biz cycle.

 

Ray's prediction:  in coming decades the issue of whether a machine can be a person,

and whether a machine can be conscious will become compelling issues

and not simply polite dinner conversation.

 

Singularity is Near came out in a 2005 book .

Ray's researchers update his charts every year

and they've stayed right on track since then.

Most recently updated in 2009.

(BTW  (my aside)  Bill Dally, former head of CSD at Stanford predicts

 300X  incr speed in computers by 2020 

due to incr in parallelism and multiprocessors. 

Bill, as CTO of Nvidia,  has NVIDIA working on it now )

 

Ray's current focus... brain and mind.... cutting edge of Singularity.

 

He claims...  brain uses 6 levels of indirection and thinking... 

1 billion pattern recognizers.  (I infer that

he's describing the operation of Mountcastle's cortical columns).

Artific. brain will not be restricted in volume to  just 1400 cc.

 

Since 1890  American census,  Moore's law is only one example  (the fifth paradigm)

it has shown extremely predictable trajectories just like the ideal gas law despite the

vagueries of entrepreneurship,  R and D creativity, innovation. 

it looks like thermodynamics..

But the law only works in the aggregate;  eg

no one could have predicted that Google would be the leader of search.

Even tho  bits produced are doubling every couple of years, 

our consumption  also more than doubles;

Also  other IT technologies are  growing even faster.

Storage capability  (mag storage, etc)  and

DNA sequencing costs.

He shows a slide that has two videos playing (DNA sequencing etc)

alongside graphs showing  Moore's Law of  DNA and mag storage.

 

Life Extension:  in ? 1800  median life expectancy was age 37.

 

BRAIN and MIND...

 

huge conceptual gap  btwn  objective world and  subject  (hard problem of david chalmers)

he agains shows image of cerebral  cortex 

(article  by John Horgan stating that brain is justtoo complex

to understand but (ironically) John used a simulation of cortex gen. by Blue Brain.

Ray was just on a panel in Israel debating with Henry Markram (of Blue Brain)

 in which Ray was the conservative...  Henry says ... will pull it off by 2018.

 Ray says No;  cannot do it til at least 2020's  (?  human parity in neural emulation ??)

 

Even if we have brain sim running, it still MUST LEARN in order to be functional...

that is the key to the brain.  It must interact with a sim. environment and body.

Consider this:  the  brain is  designed initially by the genome, which only has

800M bytes of info  and which can be  compressed down to only 50M bytes

using lossless compression.  eg  the amino acid sequence  ALU is repeated

800k times.

 

We have partially reverse-engineered the cerebellum; it is critical to

coordinating motion as in catching a fly ball;

(my note:  Prof Jennifer Raymond is working specif. on this problem at Stfd...

circuit diagram of cerebellum.)

 

 

Cerebral Cortex  (CC)  is only region of the brain where we can think

in hierarchies.  Think of something and give it a symbol -

only  CC allows us to do this.  Lizards can't do it... no CC.

CC is the size of table napkin in humans...

has one billion units .... pattern recognizers... modules...

can recognize the cross bar in a capital A regardless of shape.

Other units recognize  irony or humor (that are higher in the hierarchy of ideas.)

more abstract, but same organization.

So, we can recognize faces, pity, irony.

 

These units are organized as ordered one dimensional lists.

They are ordered and can go in only one direction; just like reciting

the alphabet - only do it forward  ABC;

 LISP  machine companies  of the seventies and eighties used same principles.

Brain also uses recursion, similar to LISP.

 

Google  is really an AI company;

narrow AI  already  generates tens of billions of revenue.

AGI will generate hundreds of billions and more.

 

AI winter in the 80's.

turns out that the LISP enthusiasts were right...

each cortical module is like a lisp statement...  incredible hierarchy..

 

we're devleoping a good idea of how these modules work.

how they are wired together.

(an axon cannot grow arbitrarily to link two modules that are  an inch or two apart)

 

Spindle cells are unique to  humans and great apes; 

We have 88,000 of them and

more of them in the R hemisphere.  Each has

100k or more connections.  They are really big cells.

They fire in response to emotionally evocative  material.

 

CONSCIOUSNESS  (Consc)...  hard to discuss because of conceptual gap..

penrose and hameroff  ... claim we can explain it...

cellular computing in microtubules, but this is a fantastic leap of faith;  consc in tubules? (NO!).

 

CONSC... Ray has never heard of any falsifiable test for consc by which

you could decide whether a device is consc; eg does it pass the Turing Test.

(that's a great challenge for consc. researchers)

 

Critics of Ray like John Searle accuse him of being a reductionist but

John favors biolog naturalism, which is also reductionist.

 

Important issue because our whole moral system and legal system depends on this.

 

Ray is writing and one chapter is called "you gotta have faith."

He equates consc and soul...  people believe this as an article of faith.

he thinks consc is not scientifically penetrable....

 he wants  a sci falsifiable hypothesis or experiment;

how do you decide whether paramecium or animals are consc or whether it's just reflexes.

 

His attitude. if it quacks like a duck then it is a duck.

When our AI characters are Turing competent, then we will believe they are consc.

 

Tommy Poggio...  visual stream is broken into 7 different movies.

(Ray uses this to reason that consc is an illusion.)

We have billions of bits streaming into our eyes.

 

Just as we now have machines that can move mountains;

we will have machines that extend our mental reach.

 

Question from Jack Sarfatti in re:  consc and quantum collapse.

 

My Question to Ray:  What in your view is the level of neuroanatomic detail required

for machines modeling the brain to be able to pull off human parity?

eg  is it  ion channels,  detailed models of post-synaptic membranes,  macromolec dynamics?

 

Ray's A:  prob. does not require any of those; just

mid-level  functional organization of circuits  (algorithms).

Ray:  important level is understanding what cortical modules do and how they are wired together.

 50k new neurosci papers every year – lots of work moving us toward a solution.

 

 

Ben Goertzel

 

BEN GOERTZEL  (CEO of Novamente and Biomind)

 

AI AGAINST AGING

(Sean McCabe introduces Ben as being the "bad ass of AI.")

 

Ben says he's glad to know he's graduated from being a pencil-neck geek

to being a bad ass.  That's progress.

 

            Usu. he talks abt  Novamente and OpenCog - his AI work.

But he's also CEO of BioMind LLC which does narrow AI applied to bioinformatics.

That is his focus today...  work done in collabor.  with Genescent.

 

            How do you define  Life?  He is inspired by Bill Clinton,

who answered  (in response to sex with Monica Q)

that depends on the meaning of the word IS.

 

Body can be understood as a machine;

Bio systems are v complex, multi-level, self-organizing.

We humans are poorly suited to analyze these data because we

evolved for hunting food,  killing enemies, etc.

 

Costs about one billion dollars to develop a new drug.

 

Why do we age?  Some issues...

 

Hayflick limit - only a few dozen replications then cells auto destruct in

a recursive cascade of self-destruction  as  dna repair  goes bad.

 

Aubrey DeGrey's approach:  as stuff goes wrong;  just fix the damage.

Biologists are skeptical - they say harder than he claims.

Aubrey wants eg to move mitochondrial dna into the nucleus of the cell,

but this will have unpredictable consequences

 (since cytochromes are inside the mitochondria).

 

Targets: Hayflick limit; dna repair;  mitochondrial damage

 

antagonistic pleotropy:  we live well as a 10 yr old, as a 20 yr old etc.

 and we have adaptations to optimize us at each difft. phase.

Those strategies are not compatible.

Death rates level off when you are  80 or 90... late life....

 annual death rate flattens out but unfort it's v high.

 

Genescent was founded by Michael Rose of UC Irvine.

They have bred drosophila  that live  5X longer than normal- months rather than weeks.

All done by selective breeding.  Now they are studying their genetics

 to find out why they live so long to

account for this spectacular increase in longevity.

 

Yes;  fruit fly life is longer by 5.5X  but this involves

about abt 1200+ genes and multiple complex pathways;

 

We applied  AI principles to this stuff.

That is the driving force behind his company Biomind LLC;

  to apply advanced (narrow)

 AI for to do postgenomic  bioinformatics.

 

eg, they are trying to work out the genetic basis for chronic fatigue syndrome;

they can do quite accurate diagnosis of genetics of Parkinsons and Alzheimers.

(no supporting evidence presented)

 

Another  bio application - use narrow AI text parsing  (Ariadne genomics or genetics)

 to extract  relationships among published papers in the literature.

 

Biomind found a few dozen key aging- assoc genes

that seem to have more importance than the others.

Look at hubs in the regulation network.

 

Some of these genes are modified by

selenium,  vit E,  estradiol,,  valproic acid,  quercetin,

calcitriol,  genistein,

resveratrol,  zinc,  folic acid,  isoflavones...

these act on proteins identified by our analyses as being  important for  longevity.

Seems to suggest that we are barking up the right tree.

 

We now have the have the snps for the long lived flies;

we are sequencing the Methuselah flies... these are rich, complex regulat networks.

 

We want to combine  number crunching with human conceptual abilities;

 

OpenCog.org is his AGI work...  two of Ben's books are the hidden pattern...

 building better minds....  will come out next yr.

 

 

***********

 

Steve Mann

 

STEVE MANN   (CS Prof at  Univ of Toronto and the original  human cyborg)

 

Extending Ourselves With Technology

 

(The mechanics of this presentation are totally cool and wow the audience.

Steve was wearing a video camera mounted on his glasses

that is connected to the internet so that  30,000 or so people see what he sees.

  He has done this for decades - see photo.

One of the massive screens that we audience members look at

shows a direct feed from his video camera.  So,  to present a diagram all he does

 is pick up a pad of paper and draw the diagram  - we see it in real time!)

 

            (I missed first few minutes)

 

What's important is the big picture over centuries- how will things unfold.

Consider  Danny Hillis and the Long Now clock;

 Steve loved this when he first heard about it -

a cuckoo comes out once every 1000 yrs. and chirps.

 

Surveillance is a clear and present danger.  Paradoxically, privacy seems to be one

of his major professional concerns  (later questioners ask him what happens to his

video camera when he goes into the bathroom.)

 

            We see potential invasions of privacy all around us.

He shows an infrared sensor on a sink that images

 your hands to make sure they get washed

and an infrared toilet  (not clear what it does). 

 

            He pioneered SOUS veillance... watch from below.

 

A camera on top of a tower watching from above would be SURveillance

  (as in the French word). SOUS veillance  (his coinage)  is human-centric

 ...  watching the watchers (eg watching officials, the police etc)

 

While he was an MIT grad student, Nick Negroponte and

Marvin Minsky were on his committee.

(Steve Mann was  building  wearable computers in high school ...

He shows us a video recording by Nick Negroponte.)

Some unnamed  MIT prof strongly objected to what

Steve does as being an invasion of privacy.

Tried to get Steve thrown out of MIT..  But Mitch Kapor was consulted.

Mitch said... wait  this is of great value.

 

Steve describes his EyeTap  device... both a camera and a display device....

it has 45 degr beam splitter... mediated reality...

laser light reconstructs what u wud be seeing.

 

He works with Martine Rothblatt... was at a future persons  confer. 

 

            (It’s great when Steve looks into the audience...

 we see ourselves thru his eyes.

He comments on the  increase in spatial frequency

of the rows of people as you look toward the rear

of this long narrow ballroom at the Hyatt.)

 

            He worked on  Interaxon  a bot-controlled display

 used in the Ontario Olympics.

 

Gordon Bell of  MSFT makes a similar camera that

 Gordon uses on his chest... looks like camera employed by HAL.

HDR  Imaging...  high dynamic range imaging....

build super hi res  construct of reality....  changes gain depending on level of light..

 

The Undigital Singularity.  He wrote an essay called Being Undigital  IEEE IST 1995

(this was a counterpoint to Nick N's   being digital...

since >1990  or so  his video feed has had open  access to internet ...

 everything he sees  30k people on the internet see and they can also "write to his retina"

He calls this GLOGGING ... blogging from sensors.

 

flUId is   flexible limitless user interface device ...  Ryan Janzen...  undigital interface.

 

Ryan is a composer of orchestral music and a cs guy... works with  Steve..

he wants to preserve that which is integral to human experience;

 

Another very cool demo -

 

The HYDRALOPHONE  this is big blue plastic fountain with water spewing out the holes.

Ryan gives this part of the talk..

Steve's Dean at Univ of Toronto loved  this  instrument

 and canceled all his  appts to play with it.

patented technology... built into a hot tub.. steve wozniak loved it.

 

he shows a roof top  hydralophone...  located at a public landmark architecture site

 

 

Ryan plays the hydralophone: a low, ethereal sound – mermerizing.

 

Here are some of Steve's websites .

 

H2Organ.com

FUNtain.ca

eyetap.org/fluid  fluid media

eyetap.org/fluid

 

He shows the ladder of life...  SOUS veillance :

Direct cameras AT high level officials  (like  CSPN does)...

we  need undersight.. 

keep eye on the officials; congressional  oversight; police, army etc.

 

Q  do u turn off your video camera in the bathroom?

Q  how abt its physiologic effects on you (as a multiyear user)

 

Martine Rothblatt...  separationof gender specific spaces...

as we age... we are gradually replaced with computer  prostheses..

may need help using toilet ... husband wife  team  use it.

A  physiol effects... what effect does it have on me?

clothing is parallel;  eg naked spa  nudist camp....

San Francisco is cold and miserable  all  yr around; 

 

Q  What do u think of theramins?

A:  He's built several theramins.  The generic problem is

that theramins have no tactile feedback.

 

*********

BRIAN LITT  BRAIN-COMPUTER Interfacing  BCI ...  present and future

 

Assoc Prof of Neurology and Assoc Prof of Bioengr  Univ of Pennsylvania

teaches an entire course on BCI

 

a second site of BCI research is the  BrainGate group  John Donahue  at Brown.

BCI figured prominently in the movie INCEPTION..

 

Reporter  asked me if we cud do AVATAR right now... LOL...  short answer is NO.

BCI  BMI  Neuroprosthetics.

BCI  includes motor prosthetics  but we need to broaden definition

Invasive stimulation in Parkinsons and in Tourettes is also  BCI. 

 

How to Classify instances of BCI:

open or closed loop   one way vs two way.

Some systems are minimally invasive...

eg  cognitive  evoked potentials  which use the  P300  from scalp electrodes.

 

next (in terms of invasiveness) would be cochlear implant... then

Deep Brain Stimulation DBS  subthalamic nucleus  stimulation for Parkinsons.

 v invasive.

finer temporal scale and spatial precision.

 

Applications:  treatment of diseases:  epilepsy  (his specialty),

depression, obesity, Parkinsons.

 

Compensate for losses of hearing, vision, gait, artificial limbs.

Restore or repair:  stroke, spinal cord trauma, peripheral nerve injury;

 

Augment:  consciousness, memory, speed, attention, perception cog processing.

This is already controversial:  Olympic   committee had to rule on

whether a below the knee amputee with prostheses could compete in regular Olympics.

 A was NO...  unfair to normals!!!

Levels of Organization:  does stimulation or monitoring occur at level of individual neurons or

with cortical columns or with  deep  nuclei. 

 

Factoid: Arthur Rubinstein could play the Minute Waltz faster than

nerve conduction velocity from motor cortex to his fingers.

Conclusion:  execution was done subcortically by over-learned reflex-like paths

that had been well-trained.

Same thing with  batter going up against a 100 mph fastball.

 Time is too short for the swing.

Batter must begin swing during the pitcher's windup.

Dr. Litt's field is epilepsy control and monitoring.

Reseaerch involves understanding  functional. networks in epilepsy (and disrupting them).

 

Parkinson's disease:  stimulate an inhibitory nucleus to upregulate  a damaged area in Pksons.

(as an aside he mentions he has 3 teen age boys... it's easier  to have pets.)

 

performance of these systems =   product of quality of info, bandwidth, and accessibility.

he shows slides from Nat Geo 2010  article see pic of child with cochlear implants...

Hearing circuits...  can interact anywhere in the system

if whole system is intact...

why would you bypass the system to go directly to cerebral cortex?

Don't !  Instead use periph processing... as in cochlear implant.

 

Feature extraction  in cochlear  implants.

people try to estimate the formants of speech f1 f2 f3...

 

Pronounce the word CHOICE : it comes out  CH OI  ssss;

decompose the word using a  bank of filters  to  extract n channels.

If u have only a few channels in one ear, the speech sounds like Darth Vader.

 hard to integrate with speech in opposite ear.

 

Alan Alda  did a Sci American Frontiers program  called Growing Up Different in

  April 6 2005  PBS  Sci Am Frontiers...

Demo'd  difference btwn  cochlear implants with

1 channel  vs 4 channel  vs 16 channel  vs 22 channel...  huge diff  as u incr # of channels.

Goes from incomprehensible noise to quite easily understandable.

can even interpret speech with your fingers resting on the speakers lips and throat.

Q why do they cut off at 22 channels

 

VISUAL Prostheses: many different designs

one design is to wear camera on eye glasses and t

hen use output to stimulate retinal ganglion cells

Kareem Zaghoul  (Neurosurg and brilliant bioengr)

will take over functiona  neurosurg at NIH .

Kareem Zaghloul made an artificial eye from

 neuromorphic chips, working with Kwabena Boahen at Stanford.

Rehab instit of Chicago...  a woman with a silicon retina

 that reproduces signals in the optic nerve...  Kareem Zaghloul..

 

John Pezarls and Clay Reid (another visual prosthes group)

 stimulate lateral geniculate to see patterns.

 

EFFERENT  2 way BCIs  (MOTOR)

Rehab Insitute of Chicago:

woman featured in Nat Geo Article  (which is online)

input from brachial plexus  output to robotic arm:

she can prepare food  inher  kitchen... see video... great video for my website.

 

Must train signal classifiers ....  to decide whether patient wants to move arm  R or Left.

Georgopolous: built classifier  to interpret  signals from motor strip.

 

Andy Schwartz Lab.....

cortical control of a prosthetic arm for self-feeding (see video of chimp eating grapes

using robotic arm)

Meel Velliste  et al...

John Donohue group   people controlling cursors on screens....

Nicolelis  monkey to robot  across country.

PLOS One 2009  speech synthesis  from cortical input...

 

Development of algorithms:  what features should we extract

 

major challenges:

biocompatibility of  tissue/electrode  interface

Brian has a huge lab with dozens of grads and post-docs. 

 

Gordon  Baltuch  U Penn colleague:    puts  cortical electrode on surface of CC...

 

lots of disease targets:  schizo depres Pkson  tourettes, epilepsy;

 

Steve Wong  is using a clustering algo to tell when u r in the right spot  (rather than listening for spikes)

 

Epilepsy  60 M  people  1/3  are resistant to meds...

 

He works on State of the Art Electrodes

12 million neurons go to 1 electrode;  5mm to 10 mm  resolution  of electrodes...

Ictal recording

Brian Litt and Gordon Baltuch

 

microseizures   r common in peole with epilepsy... small spatial extent;

sz  prob come from a cloud of  cortical columns...

so, sz may be a disorder of neuroplasticity

What we need  is  dense (fine spatial) sampling over large areas;

flexible,  irreg surface

silicon nanoribbons....

transistors on a flexible substrate...  20 microns  ... 288 channel...

a separate branch of his group is doing CV pacing and monitoring.

 

720 channel,  multiplex down  to ... small # of trodes...

high density  neural sensor...  360 trodes...

put difft sensors on  optical, chemical  etc

.Bruce McNaugton  David Euston  Science  318...

rats replay  places they have been;

fastfwd playback of recent memory sequence in prefrontal cortex  during sleep.

 

Human substantia nigra neurons encode unexpected financial rewards

 done by Kareem Zaghloul  and Justin Blanco.....  13 Mar 2009 in Science.

 

what is the future of BCI.

Izhak Fried....  idea storage;  make transfer of  knowledge, feelings, behavior

replay  salient events during sleep

 

direct  brain recording  fuels  cog sci  in TICS 2010

 

local fields  vs  single units vs eeg

cellular networks underlying human spatial naviagtion.... israeli group

 

He concludes with portrait of BCI  in 2200.

 

 

*********

 

DEMIS HASSABIS  (Univ College London neuroscientist; also video game entrepreneur)

 

Saturday  4:15 PM

 

            Machine Learning is Rapidly Discovering How the Brain Works

 

            (See his PNAS paper on Imagination and the hippocampus...  impt research.)

 

Bio vs  nonbio  AGI

 

1) Nonbio approach to AGI

 

Symbolic  AI  is the traditional way

 (formal logic,  logic networks,  lambda calc, expt systems)

FLAWS : brittle,  time consuming to train,  poor generalization.

difficult to acquire new symbols

and symbol grounding problem  eg CYC Lenat

 

most ambitious project was CYC designed by Doug Lenat.

encapsulate  all the world's knowledge  in a database.

Doug's group has done this for  25 yrs;  and total of 600 man yrs

But, when you add 1 new rule ,  its takes weeks to resolve with existing kb.

 

2) BIO systems:  use brain as blueprint

 

Consider the searchspace of possible AGI solutions:.

regime 1:  that searchspace might me small dense  .

could be a small space with a just few stars in it  where human brain is one star

relatively close together in design.

 

 

regime 2:  the space of all poss AGIs could be a large and sparse  space...

   with solns  relatively far apart and brain is only one...

he thinks  that regime 2 (this one) is the truth: a large and sparse space.

 

 

Evolution has only  produced  human intel only one time...

evidence that points to  regime 2;

ie, it is NOT easy.

 

Bio approaches to AGI...

consider a spectrum  from  abstract theory on the Left

 to bio implementation on the Right.

 

At L end of spectrum there are cognitive sci theories like SOAR (Laird/ Newell)

ACT-R  (Andersen) and OpenCog  Ben Goertzel.

Must shoehorn  new modules into each of these diagrams.  It is unprincipled and

you have  quite a proliferation of these architectures..

how do u prove that yours is best or correct.

 

At the R end of the spectrum is whole brain emulation WBE

 Eg Henry Markram  Blue Brain

or Dharmendra Modha with IBM's  Synapse  project....

big Q  is the level of detail   relevant to cognitive performance in the  emulation...

really 50+ yrs  away from doing 1 to 1 mapping of brain.

 

DEMIS favors a  system  neurosci appraoch in the middle of the spectrum

figure out the brains algorithms.

 

David Marr had 3 levels of analysis... the father of  computational neurosci.

 

According to Marr: Complex  bio systems have 3 levels.

Top level is computational.... what are the goals of the system

middle level is algorithmic:  how does system accomplish those goals:

bottom level is implementation... what is the physical realization

 

so  WBE  guys ...   want to  do the lowest level and

the cogsci  guys  are at the top/ comput level

 

he wants to be in the middle and figure out the algos..

 

there have been rapid advances in neurosci...

new experimental  techniques...

 2 photon micro,  tms,  optogenetics,  multicell record, imaging

 

how to find the nuggets of neurosci that are relevant to AGI

from the  50k+  neurosci papers in 2008

 

how to keep abreast of sea of info..

need full immersions and experience in both areas 

it and takes > 5 yrs of immersion.

 

neurosci provides  direction for agi and validation

 

classic example is  computer vision

 

 

VISION

hubel and wiesel  1959 work on  cat cortex...  2 types of cells

simple cells  tuned to preferred input stim

eg straight lines  oriented at 45 degrees toward L in primary viz cortex.

also, Complex cells   so eg   C1  has inputs  from all the S1  cells

that r oriented  45 degr  to the L 

so that they fire with a stimulus anywhere in the field.

 

HMAX model of Poggio  is bio-inspired  model...  HMAX

 he did it just right ... extracted  algorithm from

hubel and wiesel work.

 

A second example: PLACE CELLS...

rat is wandering around his cage...  trodes  stuck into its HC (hippocampus)

and researcher can  see the route of the rat because a particular cell

fires  say when the rat is in the northeast corner of the cage.

 

grid cells...  Moser et al Nature 2005....

cell firing in a regular hexagonal  grid  that tesselates personal space.

 

So,  neurosci provides direction for AGI models.

how abt validation...

 

does an algo  constitue a valid component of an AGI system

eg reinforcement learning.

 

In RL, the only teaching signal is the reward gained from the environment..

RL  = reinforcement learning.

TD learning  is a method for solving part of the RL problem ....

works by minimizing the error in expected reward.

TD  = temporal diffierence learning... sutton.

TD learning  implemented by the brain....  schultz  dayan montague in science 1997;

 

1)  no prediction and reward occurs anyway

2)  reward predicted and reward occurs

The hybrid present in dopamine neurons.

 

Identical  RL  algos

implementing RL algos as part of AGI is good... it happens in brain.

These neuroxmitrs are connected with...

dopamine... reward; serotonin mood;  ach  variability;  norepi  unexp variability.

 

hybrid approach:  (is DEMIS's approach)

combine the best of agi and neurosci

 

Example systems for emulation:

mirror neurons

model based vs model free systems

theory of mind

working mem

top down attn

control conflict  resol  by the ACC (anterior cingulate cortex)  etc.

 

CONCEPTS are the key... how to acquire knowledge in the brain;

 

3 dfft levels

symbolic  vs conceptual  (of a city eg)

perceptual  what it looks like vs

symbolic  give it aname

 

symbolic...  logic networks , symbolic systems;

perceptual :  dbn  hmax,,, htm  hawkins  hmax poggio

 

conceptual  how to do it??

 

HC-neocort  consolidation  HC = hypocampus.

HC sits at the apex of the sensory cortex

high level cortex  in assn areas.

 

HC  stores the memories of recent memories or episodes and

replays those memories during sleep at sped-up rate

gives high level neocortex samples to learn from

memories selected stochastically for replay

rewarded, emotional or salient memories are replayed more;

circumvents  the statistics  of the external  envirnoment

leads to abstraction.

 

Feynman: what I  cannot build I cannot understand.

 

 

 ***********

 

TERRY SEJNOWSKI (UCSD Prof and Salk Institute Researcher)

Saturday 5:20 PM

 

Reverse Engineering Brains Within Reach

 

He mentions to students that he used to compute

 with two sticks of wood....a slide rule!

 

He talks about reverse-engineering the brain.

 

Humans have a long history.

 

First there were marine invertebrates, then marine vertebrates,

then lands plants with reptiles, then

mammals, primates then man.

 

Bacteria have the same phosphorylation channels as us;

duplication and specialization have worked for millions of years.

 

slide:  great  picture of Francis Crick in Indian high-back chair. 

Consider that in 1 910  vitalism  was prevalent..

 

With the advent of of xray crystalography,   it was for the first time possible

 to work out 3d struc of DNA.

 

Craig Venter took nucleus out of cell and basically created transvestite  version of life.

 

Look at space/ time diagram of neuroscience techniques:

 

Space on the Y axis starting with Synapse at .0001 mm then

dendrites then neuron then layer then nucleus etc.

 

Time is on the X axis with .0001 msec and on up to seconds.

Ray K said  brain uses  10**16 calcs/sec... he agrees.

 

In this diagram, a patch clamp is at lower  L (ion channels)

  and fMRI at upper R (cm resolution and seconds)

also  PET scans.

 

Levels of of investigation: if u model the brain, then what level:  

whole cns  1m

systems  10 cm

maps  1 cm

networks  1mm

100 micron neuron

synapse  1 micron

molecules   1 angstrom.

 

When Terry was in college, total # of neurons was 10 billion;

then people found that cerebellum had perhaps 50 billion granule cells.

now  people estimate   total is 100 billion  neurons

  b/cause of  huge # of granule cells in cerebellum

 

synapse  ... abt  10 **15 synapses... THEY are doing the heavy lifting...

 

Ray K said  10**16 bits per second  bandwidth

 

bees are the champion learners among the invertebrates....

 assoc color, odor, position of flower;

Karl von Frisch... ethologist... found that out;

randolph menzel...

stimulate with sugar water and bee put outs proboscus.

one partic neuron

VUMmx1  neuron  responds directly to the sugar soln...

uses octopamine as neuroxmtr.

 

by stimualting the VUMmx1  they could condition the bee to put out its proboscus...

 

temporal difference learning  TD learning

montague, dayan sejnowski  1994 ...

nectar

P neuron is predicting the future... should you land on the yellow flower or the blue flower.

  dopamine is involved.

model could predict in detail the behavior of the bee...

bee prefers the sure thing...  100% prob of sugar

rather than occas flowers with more sugar but some with none.

octopamine in bee is like dopamine in us;

dopamine system  originates in  substantia nigra  or  VTA  ventral tegmental area

SN projects to nuceus accumbens  and to PFC  etc.

 

schult dayan montague  ? 1997  temporal differences

 

TD Gammon : Tesauro,  1994  played against  world champions...

 it really learned quite well.

see table 1...

TDG  2.1  used  1.5M training games and played against Robertie...

learned v   subtle  strategies... early blocking...

machine learning  (was not taught or designed... TD Gammon learned  it;

Deep Blue got all the publicity but TDG  was NOT hand programmed. it learned its

strategies:  it used a genl purpose architecture.

 

brain is a collection of special purpose algos.

cognitive dynamics:  can make radar more efficient

cognitive radio;  cognitive cars

see simon  haykin 2005

cognitive power grid...

can make much more efficient  grid  (plans for future to do smart allocation)

 

peter dayan

read montague

haykin  (his colleagues)

his colleagues  ? at ucsd.

 

 

*********

 

DENNIS BRAY     Prof. at Cambridge   Saturday:  5:50 PM

 

What Cells Can Do that Robots Can't

 

He is interested in wetware:  the computational capacity of every cell.

 

his loyalty is to the carbon based systems

 

he studies physiology, development and neurosci  at Univ of Cambridge.

 

Real  bio is much more complex than our models

 

he claims we really do NOT understand how humans do cognitive stuff.

 

cells are full of complic molecules

E Coli is abt 2 microns long  with lipid membrane.

has flagella;  receptors  sense the envir.

swimming, sensing,  metabolism

E Coli contains abt 10**9 molecs  made from  4300 genes.

 

Human cells  have abt 10**12 protein molecs  and made from  25k genes.

 

less than 2% of human genes code proteins.

 

hexokinase binds  atp and glucose.. spits out glucose phosphate.

 

enzymes work like logical devices:

 

feedback inhibition is universal  in bio systems;

 

membraine voltage + glutamate (and modulation by glycine)

 triggers NMDA receptor to open...

this is the underlying  mech of learning... the basis for  Hebbian synapses...

 

phosphorylation is a mech of switching an enzyme on and off...

Calmodulin  converted to CAM2 kinase...

mech of  LTP  (long term potentiation)

 

glycogen synthase is regulated by phosphorylation at 12 sites...

they can oscillate,  have memory...

proteins work togerther in small circuits...

eg 3 proteins can generate  diurnal rhythms...

Kai A B and C... forms circuit

see RUST  et al SCIENCE  318  2007.

 

small networks in cells process incoming signals

 

protein nets  in cells are highly complex

see Oda et al  in Mol Sys Bio  2005...

 

even more so when we include DNA... eg

sea urchin development  30 hrs...  v complex regul paths

Levine and Davidson  2005

 

living cells contain  v complex computing  elements....

 

muscle is a highly organized molec machine;

Consider the insect flight muscle:

actin myosin  interaction:   slide by  cleavage of atp..

 

muscles are turning over... they are not static...

 

 

**********

Eliezer Yudkowsky

 

ELIEZER YUDKOWSKY  (Singularity Fellow)

 

Simplified Humanism and Positive Futurism  :  9 AM  Sunday

 

Consider these examples of cognitive dissonance:

Death gives meaning to life...  NOT!

 

Death does NOT give meaning to life...

life gives meaning to life.

He shows Leon Kass (Bush era ethicist)   quote ..

.death is important to us humans.

death is a blessing in disguise....

 

if people had never heard of old age and death  and u  proposed the following:

let's say u can  wrinkle up, feel like crap and then die.... 

then it will really make life special.

 Would anyone actually go for it?  No Way.

 

This is just an example of people pretending to be wise.

He quotes Frank Sulloway.

as below:

 

Ninety-nine per cent of what Darwinian theory says about human behavior is so obviously true

 that we don’t give Darwin credit for it. Ironically, psychoanalysis has it over Darwinism

 precisely because its predictions are so outlandish and its explanations are so counterintuitive

 that we think, Is that really true? How radical! Freud’s ideas are so intriguing

 that people are willing to pay for them, while one of the great disadvantages of Darwinism

 is that we feel we know it already, because, in a sense, we do.

 

 

Is death good?  if u say NO then that's advice that no one would be willing to pay for.

 

Freuds ideas are so intriguing that people are willing to pay for it.

 

typical (nonsense) :  curing disease is good, unless genes are involved

vs  simplified version:   curing disease is good.

 

same  with  life and health  up to age 80 is good vs

   life and health r always better  at any age.

 

Yoda:  Shape matters not....

 

Personhood theory...  is yoda a person?... kids may say no,

Cause Yoda looks different until...

as Elie  suggests....  substitute this  question... would you EAT YODA??

 

Simplified Humanism

(

Good stuff:   Life Health Happiness Knowledge  (we want more of this)

BAD stuff:  death, sickness, pain  ignorance  (v awful, let's fix this)

 

The Way of  Chuang Tzu  translated by  Thomas Merton...

In the age when life on Earth was full  etc...

lovely sentiment but NOT TRUE... primitive peoples are not peaceable ... a la  Steve Pinker

primitive tribes were and are violent...  half of men die of homicide.

 

David Brin.:.  most societies  have nostaligia for the past golden age...

only in our society  do we have positive futurism... our best days lie ahead.

 

conjunction fallacy is when you assign  p(A AND B)  >  p(A)

adding more detail makes an event LESS probable

 even tho  it becomes psych more plausible.

that is a branch of the entertainment industry.  (that brand of futurism)

is this person telling a story abt the future or predicting the future.

 

values and goals  are not beliefs and predictions.

 

technophile:  tech is our friend, future is good, 

lets speed up progress and get there faster.

technophobe... is just  the opposite..

technovolatile.. future might be extrem good or extrem bad but isn't  likely to wind up anywhere in between.

 

 

technovolatile ... that is the position of all of us at Sing Instit ... Peter Thiel, M. Vassar, me.. all

 

Enlightenment Exuberance

 

Sir Francis Bacon in The New Atlantis... lovely quote

The end of our foundation is the knowledge of causes and the secret motion of things;

and the enlarging of the bounds of human empire, to effecting of  all things possible.

 

contrast that with Bill McKibben,  Enough (one of Bill's books)

 

"You can't stop progress.  but that's not true.  We could choose to mature"...  NO NO.

they don't have a vision of how to get there from here  (folks like Bill McK)

 

 

The New New Atlantis

 

positive futurism... climbing upward with or without a natural slope  ...

positive goals  (health, longevity)

 

3 worlds collide  (his story online)

Q abt Brave New World  (artificially induced, Utopian,  designed happiness)

A... logical fallacy  from fictional evidence... 

 

 

 

**********

Ramez Naam

 

 RAMEZ NAAM   (MoreThanHuman.Org)

 

The Digital Biome  Sunday  9:40AM

 

Ramez's slide set is here in its entirety.

 

 

We live in the Anthropocene Era... human-induced change.

 

the planet is warming...

global mean temp  is undoubtedly  anthropogenic

this is due to CO2  and methane...

will be 3 degr C warming over this century..

if we engage in biz as usu, temp will go up by 6 degr of warming...

will get 3-4 meters of sea rise... if 6 degr C  of warming...

 

(some grumbling from a skeptic in the audience)

 

This won't lead to the apes taking over  (Statue of Liberty on beach:

slide from Planet of the Apes)

 

Runaway warming is the real problem -

Increased temperature  thaws permafrost ...  releases methane and CO2...

Get feedback loops..methane is much more powerful greenhouse gas...

 and incr temp could lead to huge release.

 

there is 750  petagrams of carbon in the atmosphere...

 

plants soils have 2k PG (petagrams)

deep oceans have 40k  petagrams

coal, oil, gas have 10k...

we could get  dramatic  warming...

 

Younger Dryas ..  v rapid warming  10k yrs ago... 15 degr C warming in a few decades...

at times of scarcity,  people fight over  scarce resources... Wm Calvin  (neuro guy.)

 

CO2 gets into ocean and causes  ocean  acidification... 

corals can't create new shells, calcareous  plankton...

 

2 degr C of warming  causes corals to start to die off  and

 at > 500 ppm of CO2  they are all gone...

 

he shows data on fish tons per unit of effort...

data from Japanese fishing fleet...

they most go further and they catch less.

 

fresh water... vital to food supply...

ARAL SEA in 1989 ... vs 2009   almost gone...

Russians redireced  rivers for agric  and it’s almost gone...

 

Ogallala Aqifer: in MidWest US:  runs from S Dakota to Texas...  huge   depletion...

 

2008 spikes in food prices...  yields r not going up anymore..

He shows grain yield per hectar- Green Revolution has run its course...

 

SPECIES LOSS:  as a computer scientist, he hates data loss, loss of info:

 loss of species.. missed opportunity...

species loss is also a singularity...  rapid expansion.  not all exponentials are good.

 

World's Liquid Fuels Supply  r also  dwindling..

world oil supply is  not expanding;

 

SARS virus.. appeared out of nowhere and then  was everywhere... 

SARS models.
MRSA   (methacillin-resistant Staph Aureus) huge trajectory...

NDM-1  new superbug in UK...  bacteria do lateral gene transfer.

 

Biology as info tech

Sequencing center.looks like a data center.

 

George Church... wants to do 100k genomes  by 2020;

but he predicts  10 dollar genome by 2020  so prob we will do far more than that.

 

3 million species on earth...  why not sequence all of them

plants 290k species

insects 740k species

animals  abt 240k  species.

 

dogs....  he shows a clade  circle... beagles,  retrievers  etc..

a huge radiative adaptation ... all done over few 1000 years with selective breeding.

 

tools from biology can also be used to create

ENERGY from biology and biofuels

crops compete with food...  refining is too expensive...

craig venter...  Exxon  has paid him 600M  bucks to develop  orgs to do it..

create orgs  that do what we want by direct design... eg algae...

 

he shows a possible algae plant in nevada desert.... (artist’s  drawing)...

DARPA  wants  fuels created on-site  with biofuels  eg in Afganhistan...

Tobacco Mosaic Virus...  10 nanometer.. make a spray on PV  from them.

 

incr  photosynth  of plants  to pull CO2 from atmos.

 

Aquadvantage  salmon...  grows faster....  close to approva by FDA.

 

cocolithiphor.. makes calcium shells  30% of all calcification in oceans...

this family does More  calcific  in high CO2 conditions.

 

It is a beautiful  organism... great slides...

 

We sequenced SARS  in 5 days

 

350M  terawatt hours of energy from sun per year...

how to tap into it...

20% is best efficiency in lab with PV's.

 

desalinization... takes a lot of energy...

but not that much.

 

How about food supplies...

C3  production is used to create wheat rice

C4 production is used to create corn

 

winning the race:

 how to maximize the odds of great future...

 

optimize innovation...

 

Market assumes (tragedy of commons)  just dump  CO2  into atmos...

 no cost to you. need to price or regulate the commons

incentivize  innovation...

 

Q  gerry orstrom  (gave  10k to Sing Instit)...

how do we know CO2 is inducing warming change ?

 

A:trapping more heat in the atmos...

Q  has warming preceded CO2 yes, in some cases

But actually both r occurring...

 

 

******************

 Lance Becker

 

LANCE BECKER  MD  ER doc  and Prof. at U. Penn

 

Modifying the Boundary Between Life and Death  (Sunday:  10:40 AM)

 

life used to be simple...

alive vs dead  with nothing in btwn...

patients were rarely successfully resuscitated.

irreversible  brain damage in 4 mins.

 he suggests ... why that may not be true...

he wants to expand that bndry btnw life and death...

how big.. how long is that bndry.

 

it used to be 4 mins... how abt if it was  10 or 20  or 40 mins...

sudden cardiac arrest... arrhythmia...  95 to 98% will not be resuscitated.

 

usu causes are strokes heart attacks, brain death...

 

ischemia and reperfusion... loss of nl oxygen or blood flow...

 

what is death... we know surprisingly little about it.

when is death...

he went to the lab decades ago to find out....

when do cells actually die...

he was growing heart cells in vitro.

heart cells begin to beat together even on a glass slide.

when do the cells die when deprived of O2?.

 

heart cells on slide.... deprive the cells of O2 for 1 hour ....

during ischemic period of 1 hour the cells were NOT dying...

ONLY DURING REPERFUSION did they start to die off.

cars... run out of fuels  for 1 hr...

imagine this:  after 1 hour you put gas in the empty tank of the cars

and then they explode!!!

this is exactly what happens with neurons and many other cell types:

they only die off AFTER the O2 is restored  after 1 hr .

 

What's going on?  It appears that cell death is an ACTIVE process

it's not simply cells running out of fuel and dying.

 

He began to study the mechanisms of cell death.

all roads lead to the mitochonidria...

 

The mitochondria were key to our becoming muticellular creatures.

Billions of years ago they became incorporated into larger cells that

ate them, then developed a symbiotic reln.

 

Mitochondria consume O2 and need it to stay happy.

Lack of O22 causes eletrons to pile up inside them.

 

The reperfusion injury occurs when O2

is reintroduced into the cells and combines

with the excess electrons... that's the signal for cell death. 

 

How to prevent it?  He has tried cocktails of

antioxidants, inhibition of signaling, cooling,  mitoch inhibitors like  HS  CO CN

 

We really need a mitochond  meter (to inform us of their state).

 

Look at the article in Popular Science called COLD RELIEF .... 

cold heart can save your brain..

But cooling has a major problem...

it is v time dependent..

He developed a new slurry generation device

to cool someone immed. in the ER.

  (Cooling usu takes abt 8 hrs to cool somebody 3 degr C)

 

He cites rat research by Gerry Buckberg  UCLA... 

rats subjected to 30 mins of global ischemia...

 

uncontrolled reperfusion causes  injury... compared by Buckberg..

if uncontrolled get  2 deaths from brain herniation (cerebral swelling) in

rat control group.

 

However when reperfusion is tightly controlled (in treated group of rats)..

Buckberg got 6/6 survived;  high brain 02 uptake and

3 rats had full neuro recovery...

 

Penn Goal...  do emergency CV bypass in ER  + controlled reperfusion...

he really wants a mitochondrial meter...

This man was dead...

now he's back from the dead

Also see Popular Science.... Frozen Alive article.

 

 

******************

SHANE LEGG  (University College London, former advisee of Marcus Hutter)

 

UNIVERSAL MEASURES OF INTELLIGENCE  (Sunday:  2 PM)

 

Marcus Hutter (Swiss) has studied problem of intel for decades.

See Marcus's website on AI and on AIXI,  Marcus's genl math theory.

 

Shane starts with calcs per sec in supercomputers..

He projects out to 2020:  get  10**18  calcs per sec

We are currently at 10**15 calcs per sec.

 (Ray K's figure was 10**16 now, by IBM)

 

BUT, rather than just showing Moore's Law in terms of calcs per sec on Y axis,

what we really want is  MACHINE INTELLIGENCE on the Y axis.

Is  machine intel  going up same as Moore's Law?

 

Well, how do u define or measure intel?

slide:  Homer Simpson ...  big laugh.

 

Get def from Russell and Norvig... first few pages

 

How to approach problem:

he shows a 2 X 2 grid with

 

Human level vs ideal level   and

 

internal properties vs external behavior

 

We don't know how to do the internal properties but

we can talk about the bottom R  square in the 2 X 2 grid,

ie  IDEAL, EXTERNAL BEAHAVIOR.

 

How do we define intelligence.

He has compiled the world's largest collection of

defs of intelligence (80 definitions).

 

Def by Gudwin:  intel systems r expected to work, and

 work well in many difft environments

 

Def by Simonton: a cluster of cognitive abilities that lead to

 successful adaptation to a wide range of environments 

 

the ability of a sys to act approp in an uncertain enviro,

 where approp action incr prob of success:  Albus

 

any sys that gens adaptive behavior for a wide variety of goals... Jerry Fodor.

 

So, to summarize:  intel is a property of an agent that interacts with its enviro

to success achieve goals

across a wide range of environs.

 

Consider the following series 1,3,5,7, ???

maybe the rule is 2n -1

 

then he shows v complex eqn  where it jumps to 33.

NO... Occams  Razor...  simplest rule...  great heuristic.

 

So, to the above defintions, add in Occam's  razor...

 

Here's what you get.... he shows a small, compact closed expression for intel:

 

SUM  2**-K(mu) *  V  from mu to pi

 

pi is agent  and mu ranges over enviros

V  =  success  (reinforcement) =  pronounced  Nu.

 

E =  wide range of enviros

mu =  enviro  mu is ranging over a wide range of enviros, E.

 

Occam's razor  is the 2**-K(mu)   K is complexity;  when  complexity goes up

the weighting of that term goes down.

 

He has formally captured the definition of intelligence;

He shows that this equation captures the essence of many informal defs.

and that it orders simple agents correctly.

Upper limit is  AIXI  (Hutter's  optimal AGI algo)

can prove that it's optimal for all enviros.

It is a continuous measure

 

intelligence  spans  a huge  range;

Turing test is binary... can't guide research  pass or fail.

This eqn is non anthropocentric.

 

We want want to build the machine intelligence graph  (on the Y axis)...

r we doing better or not each year.

 

AIQ  Algorithmic  Intelligence Quotient

 

He applied the equation to a variety of theoretical systems, eg

 

Q(lambda)

 

he gets numbers  for

Q(lamdba) + functional approximators

 

MC-AIXI  Monte Carlo  AIXI

program  can learn  poker  etc.

 

if u cannot measure it, then it is not science ...  WM  Kelvin.

 

Joel Veness is his colleague.

 

Marcus Hutter  is his PhD thesis advisor (Swiss Natl Instit)

Shane is at Gatsby  Comput  Neurosci Unit

 

So, how do you actually apply it to real systems:

extremely complex to apply:  its a 50 page paper...v hard to apply

 

LEARNING is essential to intelligence:  must be able to adapt to new enviros

 andnew challenges.

 

In his testing of the eqn, he used a model system in which

 the agent pushes different buttons

to get different rewards.   He gets an audience Q:  how do u rate humans?

 

A:  don't brush off human:

toddlers r v smart...  our  algos  do not approach them in intel.

 

cannot do all poss enviros... so they do Monte Carlo sim

of all poss enviros (random sampling).

 

woman working in psychometrics asks a Q:  isn't your test order dependent?

And,  aren't some enviros far more challenging than others?  How do u rate the enviros?

 

 

**********

 John Tooby

 

 

JOHN TOOBY  (Prof of Psychology at UCSB and

 

Founder and leading light of the field of evolutionary psychology!!)

 

CAN DISCOVERING THE DESIGN PRINCIPLES OF NATURAL INTELLIGENCE

UNLEASH BREAKTHROUGHS IN AI?

 

Sean introduces him: "Evol. psych has shown that sex evolved as a

defense against parasites.  And here I thought that it evolved just for fun."

 

His colleague at UCSB is Leda Cosmides. 

 

In regard to future of AI, he thinks that

truth may be really far from what we expect

rather than lying btwn the extremes defined by

the AI  skeptics  vs the AI  true believers.

 

His goal in evolutionary psych is to reverse engr the code of human nature...

 

object of study of evol. psych is the

neurally implemented programs that evolved to regulate behavior...

 

there are two different ways to think abt it  ( 2 difft Marr's levels)

1) the physical...  Marr's implementation level... neural strucs  etc

2)  thecomputational  (Marr's algorithmic level)

reverse engr  the info-processsing ops and data strucs.

 

why emphasize evolution  (the black art)?

 

natural selectioon  acts as a filter  on bio struc:

  it only builds func struc that solve  evolutionarily  recurrent adaptive  problems.

 

human brain consists of a collection of problem-solving strucs...

Einstein quote:  it is theory that decides what we can observe 

(unguided empiricism is too weak and slow);

 

ancestral world posed a series of problems  that

had to be solved to survive and reproduce...

naviagtion, mating, foraging, negotiation in conflicts of interest  etc.

 

the architecture of the adaptive solution  (the circuit logic of the program)

 reflects ...in detailed ways,  the structure of the adaptive problem...

they mesh like a lock and key..

 

huge amt of adaptive structure... enormous diversity outcomes....

beautiful slide showing the  huge diversity of animals.

 

 

Here's the EVOL PSYCH  5 step program.

 

1)  ID  an enduring adaptive problem  faced by our hunter-gatherer  ancestors

2) do task analysis  to hypthesize abt poss soln programs

3) test hypthesis in lab

4)  id the program's  neural, developmental and genetic basis

5)  test  cross culturally

(he uses a field site in Ecuadorian Amazon)

 

He thought:  psych  MUST be using evolution...

he took  a course in cog  psych in 1970 and he was

flabbergasted to find  no  use of evolution in psychology..

 

Still, even in 2010 evolution is eft out of cog sci, psych, neurosci.

why would anyone leave it out...

like making a taboo of math.

 

He quotes Max Planck: Science  progresses funeral by funeral!

(leaders of the field have to die off and be replaced)

 

Example: (Darwin was stumped by altruism... an apparent contradiction to evolution)

So,  what are the computation algos for evolved programs for detecting

 genetic  relatedness, and using that to explain altruism...

 

  1. need to avoid  mating with close relatives....
  2.  to avoid getting deleterious  recessives  ...

selects for  avoiding mating with relatives..

 

2) BUT want to care for and aid close genetic relatives (including offspring)

and to avoid harming them

 

Kin selection  Hamilton's rule... help relatives...

solving these 2 difft probs:  ancestral adaptive probs, required the same solutions

You need a kinship index...  need that..

needs to affect the motivational system causing sexual attraction...

you  can't neces. see relatedness..

what r the cues to relatedness...

domain  general learning is not enuf. (need specific expertise)

 

evolution sifts for those cues which r v reliable  generation after generation

 

Cue:  maternal  perinatal assoc  MPA... 

kid observes  mother  nursing  a new baby then baby is related to it.

 

trying to explain incest aversion..

when hunter-gather  bands  fission and fuse,  nuclear families stay together...

degree of kinship for sibs  is related to length of residence....   goes up to age 18...

 

convergin evidence...

cue monitoring system.

co-residence monitoring  system..

 

intel is achieve by throwing away  irrelevant info...

if u have  maternal perinatal assoc  MPA  but if you're  a younger sib and

u lack this then use  co-residence  monitoring system...

 

kinship index  down-regulates sexual interest...  leads to disgust.

 

these mechanisms are nonconscious:  computation of KI  (kinship index)

explicit beleiefs abt kinship do not matter 

(it is the nonconsc piece that determines their attitudes)

 

people encode race because  it's related to cues abt coalitional alliance...

 

We really need strong AI to understand what humans want.. 

must have  maps of  of human mind ...

human... what is the optimal  population for sustainability...

 

Rather than AGI (general intel),  Tooby favors  the term BROAD intel...

ie it's a matter of degree  .

 

What are the hacks that we use to accomplish our goals?

biomimicry

 

intel is computational  organization that solves problems.

 

dedicated  intel  uses highly target problem sets

 (uses  nonstandard  adaptive logics)

 

improved  intel  improvises solns to probs not encountered before...

 

useful intel is all abt reduction  because

 

all intelligence  is too bulky to fit in our brains or our servers...

what r ways we can massively carve away the stuff we do not need.

 

He states: goal of problem independent solutions has hampered the development of AI.

 

rather than that: it's  better to find pinpoint strategies that solve  key problem sets..

color constancy, social xchange,  parallax..

native intels r dramatically better... special purpose...

 

we were faced with narrow probs... predator attack,  foraging, social  interax.

 

genl methods r computationally weak, and the more genl,  the weaker!!!

 

Ecological  rationality...  ecolog struc is enduring...

Goal is to  develop an ever larger battery of dedicated intels

bundle together dedicated intels...

food eval,  kinematic geolocation,   mate eval,  face processing, etc.

 

My public Q:  Where can we find out more abt Evol Psych.

A:  go to his website at UCSB.

 

my second Q:  How do we know that the answers supplied by

 the field of Evol Psych arenot simply post-hoc  "just so"  stories ...

 how can we validate them?

A:  many times, when we look for a soln, 

what we find (eg with incest taboo or kinship index) is

something unconscious that noone suspected beforehand.

 Results are surprising and

then need to be cross-validated across cultures.

 

 

*********

 Zeno

 

 

DAVID HANSON (CEO of Hanson Robotics)

 

EMOTIONALLY INTELLIGENT MACHINES

 

(Sunday:  4:25 PM)

 

Hanson robotics:  we bring robots to life... 

example:  model of Einstein... 

patented skin material with improved facial physiol.

Zeno:  a consumer robot product

 

(Davbid was an Imagineer...at Disney)

 

We love characters...

characters who understand us and see us.

 

His main message:  development of intelligenct characters are a

key driving factor in AGI  (and especially friendly AGI, since they

are forced to interact with humans by design)

Example:  Zeno:  Hero from Tomorrow.

 

His Robokids are analogous to trditional character media.

 

He shows a conversation between Amanda (a human) and Zeno (robot)

this work pushes the frontiers of machine perception,

 machine learning,  material sci,  robotics

He shows an android portrait of Philip K Dick (the great sci fi writer)

 VALIS  vastus  active?

face tracking  2005...

 

We fall in love with the characters in our favorite novels:

life as a quest...  Zeno of Allea...

Computational models of humans:

 we r hard-wired for social cognition and face to face interaction.

 

We want to embody social cognition...

We want our robots to seem like protagonists.

people innately favor  agency or characters;

people desire smart, lovable chars. (eg Bugs Bunny)

push friendlier AI into mktplace.

modeling people helps us to understand people better

robotics, software, neurosci  r loops

we will discover the answer to the question: what is human?

 

defining character... means agency... beleiving that there is a being there.

A character has a story with accomplishments and goals.

A moral character: able to get along with people and to find solns.

 

Our best charcters achieve their goals by honor and by perseverance.

even newborns respond to faces... facial expr r universal ...

Paul Ekman  1972,  and 19th cent.Darwin.

 

The age of character machines is coming upon us:

Sims,,  world of warcraft,  Club Penguin,  Zwinky,  Webkinz...

 

Consider Brazeal/Winston   Kismet.

CMU Doc Beardsley

Repliee:  Ishiguro  Lab

 

Several innovations from Hanson Labs:

in robots, poor materials are too stiff and the robots are power hungry.

omposs to make mobile, walking robots.

He shows videos of Zeno.

 

At Hanson, we invented FRUBBER:

 lipid bilayer  elastomer...  lifelike and low power to deform (on the face of a robot)

we want emotional connection.

by deploying these robots as products we get a lot of customer feedback

This work combines many fields: neurosci,  materials, art,  robotics,  psych...

frubber ...  better expressiveness , using 23X  less power...

extreme pore packing density...

electric field sensing

 

Character (AI) engine:  he shows an interesting slide 

 (get this from his work at Hanson Robotics) 

Speech reco  uses Dragon Natl Speaking

Facial recog and expression characterization:

Kino Coursey  and Daxtron Labs...

Apollo Mind Initiative... want > human  intel by end of decade...

facial express mimcry..

See David's TED talk...  audio localization...

 

in development:  sentience engine...

ART is in the loop also.

bold new artforms... reifys  AI...

BINA  rothblatt...

robots can be used for autism research...

want mkt pressure for friendly  AI...

robotic character development is a clear path to friendly AI,

 

 

******

JOSE LUIS CORDEIRO  ( Director of Millennium Project in Venezuela)

 

The FUTURE OF ENERGY  (Sunday: 5 PM)

 

 

He graduated fom MIT.

 

The largest industry on earth is the energy industy:  8 Trillion dollars per year.

See the report by the millennium-project (online)

endorsed by Sec Genl Ban-Ki Moon.

 

Cute slide of Ostrich using a jet pack to hover.

 

He was at Singapore University.

worked on Home Energy Generation:

Larry Page (Google fndr) says this is what I would want to do if I was a student.

Ray K and Larry Page...  how to make solar energy economic for  humanity...

 

solar flux is  86k  terawatts per year.

hydro  power is tiny

peak oil will occur in the 21st century

Sheik Yamani,  2000  Saudi Arabia...

peak whale oil in the 19th century.

sun is approaching a solar maximum;

Club of Rome  (Jay Forester)  that report had incredible impact but was terribly wrong.

Herman Kahn...claims no limits to growth...

there is plenty of oil he says...

Venezuela...  oil for 4 centuries...

The big problem is not that oil is not abundant but that it is dirty:  buring it gens CO2.

 

Different waves of energy use:

wood in 19th cent  then coal, then oil, then natl gas,  then  other...

MIT energy initiative...

Wireless electricity xmission... MIT

Arthur Clarke:  any advanced tech is indistinguishable from magic....

 

Bob Metcalfe (Fndr of 3Com)  The ENERNET...

wants entrepreneurs and  prosumers to control renewable,  distributed  systems

 

Exxon gave 600M  bucks to to Craig Venter..

Einstein showed that mass contains the equivalent of 90M  megajoules per kgm.

 

only 15 people in our audience had heard of Kardashev... Russian  physicist.

 

in 1 hr we get as much solar energy as  we use in 1 yr....

 

he wants to reach Kardashev level 1...

 

JAXA (Japan Space Agency) want to power Tokyo  by solar power by  2030  from space...

could use moon-based panels.

 

funny slide:  we  SHELL not  ARCO...

sunny slide the American view of the world:

(summarizes S America as simply a supplier of  cocaine and coffee...)

we may ahve free energy in 30 yrs from solar.

he is moving to Korea to Seoul.

He shows an incredible picture of the new designed city that he is moving to in Korea.

The Yin and the Yang:  each has some of the other inside.

 

New Songdo City, S Korea....

Chinese saying:  in crisis there is opportunity.

 

see his site cordeiro.org

His slides r available there.