Tip:
Highlight text to annotate it
X
BRAD TEMPLETON: Welcome to Authors at Google.
I'm Brad Templeton.
We're here today with Mez Naam, who has written a new
book called Nexus, nothing to do with Google.
Yet his talk is going to be about Google and your brain.
Mez used to be at Microsoft, where he was a product manager
for Microsoft Outlook.
How many people remember Microsoft Outlook?
And also instrumental on Internet Explorer.
But of late he's become a light in the transhumanist
field, writing books about what happens when new
technologies actually change what it means to be human.
He has three books coming out this year.
One of them is a science-- two of them, actually, are science
fiction novels, Nexus being the science fiction novel as
well as a nonfiction novel.
He's here to tell us about it and some of the ideas that
he's been promoting in this field.
He also is a guest lecturer at Singularity University, where
I'm also present, and has been working on changing the world
for a long time now.
Mez.
[APPLAUSE]
RAMEZ NAAM: Awesome.
All right, thank you, Brad.
Thanks Matt Cutts as well, for you guys for setting this up.
Yeah, so I have a book out called Nexus.
I hear that a company, a technology company in the
Valley has a similar product or a product
with the same name.
I'm waiting for the co-marketing dollars.
So just contact me if you're watching this.
This is a science fiction novel set in the near future,
2040, which sounds like a ways in the future.
But it's as far ahead of us as 1985 was behind us.
That's how close it is.
The premise is that there is a drug that you can take that's
packaged as a drug.
But it's actually a set of nanodevices that attach to
your brain and transmit wirelessly what
your brain is doing.
And so if two or more people have this, their brains sort
of sync up.
And you get weak telepathy, is what Corey Doctorow
described it as.
And my protagonists are working on enhancing this.
They live in San Francisco.
And there are various battles between a US government agency
that wants it illegal and a foreign power that may or may
not be using it for mind control.
So it's a thriller.
it's really fun.
The science in it is as real as I can make it.
Most of the science in it I researched for a book I wrote
in 2005 called More Than Human about the actual science of
augmenting our minds and bodies.
Now of course, we've seen this idea of getting data in and
out of the brain in science fiction for a long time.
We've seen in the Borg in Star Trek, for instance, where
connectivity of the brain makes a super-organism that's
very homogeneous.
We've seen it in The Nexus, where the great American
philosopher, Keanu Reeves, gets to say I know Kung Fu.
BRAD TEMPLETON: That was actually called The Matrix.
They haven't renamed it.
RAMEZ NAAM: Oh, sorry.
This film with this guy.
But we've also seen some people talking about it in
other contexts.
For instance, a certain technologist here in Silicon
Valley said you can imagine your brain being
augmented by Google.
You'd think about the question and Google would whisper an
answer into your ear.
Or maybe Google could just implant the
answer in your brain.
Sergey has said we'd like to think about Google being the
third half of your brain.
So I kind of like that idea of having 1.5 brains.
You could have a bumper sticker on your car that says
"my other brain is a data center." As we all know,
Sergey is well on his way to being an
augmented transhuman already.
So this leads to the question of, could you have Google in
your brain?
Well I'm a program manager.
So I think about, what the requirements for this product
if we want to build it?
Well, you want to have five things, I think.
Data input, you've got to get data into your brain.
Data output, you've got to get data out of your brain and
into the machine or the cloud or other people's brains.
We're going to have to do some encoding and decoding work,
because the format for this data is probably not the same
between our machines and our brains currently.
So we're going to have some converters in there, some
protocol shifting.
You want multiple data types, at minimum, text, audio, and
video, and hopefully some other data types as well.
And of course it has to be safe, secure, and deployable,
private as well, which might actually turn out to be the
largest challenges of this technology.
So of course that sounds like science fiction.
But what I'm going to tell you is that very, very, very early
versions of this are possible now, that we have low
bandwidth, low fidelity transport of various data
types in and out of the human brain right now, and that it
is getting better very, very quickly.
Now why is it getting better.
Who's investing in this?
Is this a Google X project?
Mostly it's not.
It's really being developed for medical reasons.
There's a large number of deaf people in the world, 20
million in the US who are hearing impaired.
Two million of them can't be helped by normal hearing aids.
They have what's called sensorineural deafness.
That means the inner ear hair cells are damaged or
completely gone.
And so they cannot turn sound waves, vibrations in the air,
into nerve impulses.
There's 1.3 million blind people in the US.
There's two million paralyzed people in the US alone.
1 and 1/2 million people with Parkinson's disease, which is
essentially not exactly untreatable, but
uncurable right now.
5 and 1/2 million people with a traumatic brain
injury of some sort.
This is a scan of the brain of Phineas Gage, a very famous
patient you had a railroad spike
launched through his head.
He lived through the process but came out a very different
guy with a different personality, and modifications
to his attention and memory and so on,
and not for the better.
So at least 10 million people in the US who are candidates
for this technology, and really probably close to a
quarter billion people in the world who suffer from these
problems as well.
So that's the motive.
So how does this work?
Well, there's a long history of this technology.
It goes back to 1870 when two German physicians had this
crazy idea of an experiment to show that the nervous system
was electrical.
It was so crazy that the university, the University of
Berlin, , wouldn't allow them to perform it.
So instead they used Hitzig's dining room table.
They put a dog to sleep with anesthesia.
And then they electrically stimulated parts of its brain
and found that they could cause its
paws to move on demand.
So that was the first indication that the nervous
system is electrical.
We really got the first proof that it's not just muscle
motion but actual cognition is electrochemical in the 1950s
thanks to this man, Wilder Penfield.
He was a neurosurgeon who worked on a number of things.
But among them was surgery for epileptics.
Back in that day, and continuing to today, one of
the techniques used for epilepsy was called oblation.
What that meant was you did brain surgery, found the part
| the brain that was sending out these incorrect signals
leading to epilepsy, and you cut it out.
Now Penfield wanted to minimize the
damage to his patients.
So he would electrically probe the brain
while they were awake.
When he'd find the part of the brain that was responsible for
the seizures, they would have a seizure.
So then he would know to cut out just that
spot and nothing else.
But when he simulated other parts of the brain, he'd find
that people would suddenly have a waking dream, a vivid
smell, a vivid insight, some emotion.
So that was the first conclusive demonstration that
thinking and feeling and your senses are electrical as well.
From there the story gets very weird.
In the 1960s, two scientists, Jose Delgado and Robert Heath,
one of them, in fact, funded by the CIA, did a variety of
experiments on animals and then on humans to show that
they could control behavior through electrical
stimulation.
With electrodes implanted in the brain, they could put
animals to sleep or wake them up.
They could make an animal hungry or satiated, aggressive
or fearful.
With one experiment in a troop of monkeys, they could change
which monkey was the alpha male of the troop by
stimulating the right parts of the brain.
Delgado, a Spanish scientist, was quite a showman.
So he repeatedly did this experiment.
This is him in a bull ring in Cordoba.
Delgado is the guy on the right.
He's got a cape.
And you could just barely see he's got something with an
antenna here.
The bull has been worked up by someone else before.
So the bull starts to lean forward.
It's getting aggressive.
You could see more clearly, A, there's something on the
bull's head, and B, this black box with a long antenna.
The bull starts to charge him.
Delgado drops the cape and hits the button on this radio
device that he has.
And the bull skids to a halt because it's now terrified of
him, because he has remotely stimulated an input of
electricity to its fear centers in its brain.
I'm going to show you a video of this.
This video is from the French documentary in the 1960s.
So it's in French.
It's not exactly HD.
Delgado is the guy in white who appears a
little bit into it.
And watch how the bull reacts to him at various times.
[VIDEO PLAYBACK]
RAMEZ NAAM: That's the bull running away from him.
He hits the button.
[END VIDEO PLAYBACK]
RAMEZ NAAM: So Delgado went on to write a book called
Physical Control of the Mind, Towards a Psychocivilized
Society, where he posited that we could improve society
dramatically by fitting criminals with brain implants
that would cut off their criminal urges, so on and so
on and so forth.
The medical community and scientific community's
response to this was that it was
extremely, extremely creepy.
And can we please stop all that and head on to some
medical applications of this tech, which they did.
There's a TV show called House with a doctor called House
that solves all sorts of crimes-- or all sorts of
medical riddles, I think.
Well the original Dr. House, Dr. William House, was the
first person to find a real medical use for this
technology.
He invented the cochlear implant.
It looks like a hearing aid, but it's for those people that
have no hair cells in the inner ear.
So no hearing aid can help them.
Instead what it does is it picks up sound of the
microphone.
It has a small compute that does a re-encoding, if you
will, from vibrations into nerve impulses.
And then it electrically stimulates the nerves of the
inner ear, the auditory nerve fiber.
And that causes people to hear.
The auditory nerve has 30,000 neurons in it on each side.
And the first cochlear implant had one electrode.
The best ones implanted in people today have 23.
So you wouldn't think it could work at all.
But it actually produces hearing that is not as good as
a normal person's, but is enough that they can read--
can hear what people are saying and interpret it
without reading lips.
And it has been life changing.
About 200,000 people around the world have this today.
This is the picture an eight-year-old girl hearing
for the first time.
And I'm going to show you a video of an eight-month-old
boy who has just had his cochlear implant
implanted and activated.
[VIDEO PLAYBACK]
-Here we go, turn it back on.
And he's back on again.
See how he turned?
-Hi Jonathan.
Hi.
Can you hear that?
Hi sweetie.
Could you hear that?
[LAUGHTER]
-Hi.
Hi.
-You've got that, Deb, right?
-Hi Jonathan.
Hi.
-We call that a late Christmas present.
RAMEZ NAAM: So this is a life changing technology.
Like I said, 200,000 people around the world have it now.
[END VIDEO PLAYBACK]
RAMEZ NAAM: This little boy turned five in December and
he's still doing great.
Now I told you along the requirements, we need to have
data input into the brain.
So I've just shown you proof of principal of data input of
one data type, audio.
Now we all care about our hearing.
It's very important to us.
Another sense we care about a fair bit is sight.
So let's see data input for sight.
This man, Jens Naumann, this picture was on the cover of
Wired magazine in 2002.
Jens, when he was 18 years old, worked on a railway.
And one day with his pick he struck
something he got was rock.
It was steel.
A little fragment came up and destroyed one of his eyes.
Lives in Canada, very active fellow, refused to compromise
about his lifestyle.
The next year he was out snowmobiling.
And part of the clutch had a problem and threw a fragment
of metal that destroyed his other eye.
And he was rendered blind in both eyes.
He was blind for 20 years, until 2002 when this happened.
So what he's wearing is a CCD camera, not that different
from any digital camera today.
The cable here goes to a small computer that he wore on his
hip, actually it was about this big back then.
And now it's probably the size of a quarter.
And from there the signal is translated from the format of
the camera to one that his brain can understand.
And it travels back up into a jack in the back of his head,
Matrix style.
Jens has this system which has 256 electrodes, as opposed to
the many billions of neurons in his primary visual cortex.
So again, there's a huge mismatch in the size of the
pipe to the size of the system, if you will.
You wouldn't think it could possibly work.
But if produces what they call limited mobility vision.
What does limited mobility vision mean?
Well, here's Jens.
[VIDEO PLAYBACK]
-I was able to very carefully drive and look from my left
side of my right side, making sure I was between the row of
trees on the right and the building on the left.
And when I got near any obstruction in the front, I
would see that there was an obstruction.
I would also see the lack of obstructions.
And went I backed up, I would be able to inspect for
instructions there.
It was really a nice feeling.
[END VIDEO PLAYBACK]
RAMEZ NAAM: It was really a nice feeling.
So that's a guy who was blind for 20 years driving a Mustang
convertible.
Now Jens has terrible vision.
It's really, really atrocious.
I wasn't in that parking lot with him and you probably
wouldn't want to be.
But it's a big step up from zero vision whatsoever.
So that's data input.
And throughout--
in the novel, you see examples of people getting video and
audio going straight to their brain.
But what about data output?
Can we get output out of the brain to control
systems or send data?
Well we can and we have.
So this is a doctor named Phil Kennedy and a patient name
Johnny Ray.
Johnny Ray was 52 years old, a Vietnam vet, drywall
contractor, played blues guitar.
One day he had a massive stroke.
He woke up in the VA hospital in Atlanta, paralyzed
from his neck down.
And to save his life they had to do an emergency
tracheotomy.
So he couldn't speak either.
His only way to communicate was to blink his eyes once for
a no and twice for yes.
That was it.
This guy Kennedy had been working in monkeys and had
demonstrated that he could put a single electrode in a
monkey's motor cortex, the part of their brain
responsible for motion, and train them to move a cursor
around on a screen.
He applied to the FDA for permission to do it in a human
patient for the first time.
They said yes, with some conditions.
One of their conditions was that he had to seal
the skull back up.
They didn't want wires running in and out of the skull and
having a risk of an infection getting into the brain.
So they made the system wireless.
It actually sits inside of his brain.
It gets power and data from a cap that he wears, or from the
equipment near his hospital bed.
Now this is the most extreme case of this mismatch.
Because it's literally one electrode in Jonny Ray's
brain, but more than a million neurons in his motor cortex.
But that gave him enough control to be able to type out
messages on an on-screen keyboard, which is a dramatic
step up from just being able to blink at people.
And this technology has gotten better and better, and is in
more serious human trials now.
And in fact, a video came out just a couple months ago of a
woman using this to do something tat was very, very,
very important to her, which was to
have her morning coffee.
[VIDEO PLAYBACK]
-You're watching the most advanced brain machine
interface in action.
Kathy Hutchinson is paralyzed and unable to speak.
But just by thinking, she's able to control the movements
of this robotic arm, and drink her morning coffee.
She's part of a pioneering study run by researchers at
Brown University in the US.
-People who are paralyzed have their brain disconnected from
their body.
So they're not able to--
[END VIDEO PLAYBACK]
RAMEZ NAAM: So we've seen data input now of video and audio.
We've seen data output of motion.
And now in primates they have data input as well.
They have systems like this that also sent touch data back
into the brain.
There's one other domain that I want to talk about, which is
Parkinson's, that I mentioned early on.
All Parkinson's drugs fail over time.
They just stop working eventually.
But now we're finding that with something called a deep
brain stimulator, a single electrode run very deep into
the brain, you can keep a Parkinson's patient healthy
and functional for quite a while.
And there's 80,000 of these that have been
implanted so far.
But interestingly enough, they found some side
effects with these.
In some patients who got a deep brain simulator who had
previously untreatable depression, that wouldn't
respond to any drugs, they suddenly had normal mood.
In some patients that had previously untreatable OCD,
they found that suddenly they had normal behavioral control.
And so now there are clinical trials going on of using this
deep brain simulation for depression and for OCD, both
of which seem to come from very deep, old
portions of our brain.
Which, if you think about it, is sort of the same sort of
work that Jose Delgado was doing in his animal control,
but in a much more benign, therapeutic sort of context.
So that's what's happening so far in humans.
But what are the opportunities?
And I did talk about this being the science behind a
science fiction novel.
So it's not going to stop at just this sort of thing.
The opportunities are really in the higher functions, if
you will, memory, attention, learning, the things that make
us really human.
In the film Memento, Guy Pierce played
the character Lenny.
And Lenny can't form any new long-term memories.
He will see something.
He'll learn it.
He understands it.
But within a few minutes, it's gone from his memory.
This is an extreme case, but cases like this and many, many
milder cases of this sort exist in people who have
damage to part of their brain called the hippocampus.
If your hippocampus is damaged, you can't form these
new memories.
So working on this, scientists have been trying to create an
artificial hippocampus, a hippocampus chip that is
structured like the tissue in the hippocampus, and that you
can shunt in.
So in a rat experiments, they actually take a rat that has a
damaged hippocampus, this area, and they put in
bypasses, arrays of electrodes basically, that put in place
the hippocampus chip as a replacement for the damaged
brain tissue.
And they found that they can restore these rats' ability to
learn new things, restore their memory, which is
awesome, has great application for humans.
But they went farther than that.
And they've show that they can improve these rats' memories.
They can improve the speed at which they learn.
They can keep the memories around indefinitely because
there's no need for the memory in a chip to decay.
And they can reactivate that memory at will.
They can put the rat back in a maze that it's learned and
reactivate the memory of that maze's layout and have it have
perfect recall what was going on inside of that and how to
find the water table or the water platform at the end.
So that's the early glimmers of work that would actually
potentially repair, and maybe even improve,
human higher functions.
We've also gone into what we would think of as
intelligence.
So in rhesus monkeys, people have been working on a chip
that goes into the frontal cortex, the most executive
part of your brain that controls attention and
thoughts and juggling various different things are going on.
So they take these rhesus monkeys.
They put this implant in their frontal cortex.
And they give them a monkey IQ test, if you will.
It's called a pick and match test where they have to learn
to match up certain things together.
And while this is happening, the implant learns the pattern
of what goes on in the neurons of their frontal cortex when
they get it right and when they get it wrong.
So then they take these monkeys and they impair them
to get them to have a lower score on the test.
They impair them by giving them large doses of ***.
So the monkeys aren't complaining.
They're pretty happy with the experiment, I think, but their
scores actually do go down.
Then in half the monkeys, they put the implant into an active
mode to see if it can restore their skill.
And it can.
But it also goes farther than that.
The average score of the group that was not impaired or
enhanced was 75.
The group that had the implant put into in active mode scored
on average an 85.
So they're actually using this implant to improve the
monkey's skill on this monkey IQ test, if you will, which of
course will lead to the cyborg planet of the apes.
But the real frontier, I think, isn't so much in
enhancing individual capabilities, though that is
something that we will actually look at.
It's really in communication.
Once upon a time, computers were about solo work.
You had a computer because it was a fancy word processor,
which really meant it was a fancy typewriter.
It was a fancy spreadsheet machine, which really meant
that it was a very fancy abacus, in a certain sense.
But that's not what actually changed
the world with computers.
What's changed the world has been their ability to
communicate things between us, and with the cloud overall.
And that's the direction that I think is most exciting about
this sort of technology and the direction that we really
go into in the novel.
There is some work happening on that today.
DARPA is very interested in all sorts of things that
improve human performance.
And they have a particular program called the Enhanced
Battlefield Communication Program.
And that program gave a grant that funded this study.
This study had two monkeys in different rooms, with
soundproofing between them, wired together via implants in
their auditory cortices, so the
auditory part of the brain.
They would play a sound for monkey one.
And what they found as they hoped was that monkey two
could hear that sound and could identify it, could say
what it was.
And DARPA's dream is to build implants for soldiers where
the soldiers all have intuitive knowledge of what's
happening to the other soldiers in their squad and
the battlefield more broadly.
Real science happening, not yet in humans.
We've also recently--
so now we've got motion out of the brain.
We've got audio out of the brain.
This is video out of the brain.
In humans--
this study was done last year.
They put subjects in an fMRI machine, a brain scanner.
It showed them a clip from video.
And then an algorithm tried to match that clip to what they
were actually seeing.
So tried to match their brain activity to what they were
actually seeing.
Because it's techies here, I can talk about it
a little bit more.
What they actually did was the algorithm had a library of
fuzzed out video frames that it could access, and a machine
learning model that learned how to-- or was trained to
match up different patterns of brain activity with the most
likely clip from its video library.
And let's see how it actually did.
Now this is very, very early technology.
But you can see that just seeing what's happening in the
patient's brain from the outside, it gets some vague
idea of what it is that they're looking at.
Which is exactly what we'd expect.
In general in the brain, the same neurons fire, the same
activity happens, when you're seeing something, when you're
imagining that thing, and when you're remembering that thing.
There's some difference in higher order bits of the
brain, but in the primary visual areas or the primary
auditory areas, it's the same sort of thing.
All of this, by the way, happened with brain scanners
on the outside.
And those are very, very limited in how high the
resolution they can get is of what's going
on inside the brain.
With actual electrodes in the brain, this quality would
probably much, much higher.
And now the team funded by DARPA that did the hippocampus
improvement for the rats is looking at linking the
hippocampi of two rats together and seeing what could
happen, seeing if one rat who's been trained to run the
maze can convey that information directly to
another rat that's never seen the same maze.
Now like all experiments, the first ones of this will almost
certainly fail, but we will eventually get there.
That's the kind of thing that people are working on now.
Then that leads, of course, to Keanu Reeves
saying, I know kung fu.
Now let's talk practicality.
I've been spinning out what this sci-fi world could look
like and it's really fun to play with, but there are some
real limitations.
The first limitation is quality.
Can any of you say what this image is on the right?
If you've seen this example before, please don't say.
Anybody who hasn't seen it, can you pick this out?
Is it Jesus?
Is it the Statue of Liberty?
It could be both.
AUDIENCE: Woman with a hat.
RAMEZ NAAM: Woman with a hat.
Very good.
That's awesome.
AUDIENCE: [INAUDIBLE] put a hat on the Statue of Liberty.
RAMEZ NAAM: On Jesus.
So Jens Naumann, the blind guy, he has 256 electrodes.
He has a 16 by 16 pixel display essentially.
This is the quality of his vision.
So you can see why nobody wanted to be in that parking
lot when he was driving in the car.
He can basically see large obstructions and not
much more than that.
So quality is a big limitation.
And indeed, if you look at the most sophisticated brain
computer interface we've ever implanted in someone, has this
256 electrode situation right now.
And that's largely because the people who have been doing
this have been neurologists and neuroscientists and not
electrical engineers.
That's not who've been doing it so far.
But they are starting to get into it.
And now you see labs devoted entirely to developing the
hardware that's being used in these experiments.
University of Michigan has 1,000 electrode wireless chip.
And a German company called Infineon has a 16,000
electrode wireless chip that's coming on the
market right now.
So we will see that bandwidth go up.
And there's huge room to improve.
The chip in my laptop has about a billion transistors in
it in a size not much different than some of these
interfaces.
Of course, these things are very hot, but we don't have to
run them at gigahertz.
If you run them at 100 Hertz, your going to pretty much
match what the brain can do.
So it's quite possible with advanced technologies that we
already have in other domains to dramatically increase the
bandwidth coming in and out of the human
brain in these systems.
And we don't have to go far for it to suddenly look pretty
functional in terms of restoring functionality for
someone who's been impaired.
At 5,000 pixels, this is what you see, which is a dramatic,
dramatic step up from 256.
And 5,000 pixels is still dramatically smaller than the
type of technology that we use in electronics traditionally.
The other very big limitation is this is your brain we're
talking about.
Very few of us are going to voluntarily have brain
surgery, certainly not to get 256 pixel vision, or even the
5,000 pixel vision.
So this is a very big hurdle.
In the book, I get around this by saying, hey,
you swallow the drug.
The nanoparticles go through your blood-brain barrier and
self assemble around your neurons.
It's all great.
But that's not the technology that we have today.
Of course there have been other cases where we've gone
from something being so unsafe and so impractical to being
very practical.
They used to say that eye surgery would
never really take off.
Our eyes are pretty precious to us.
Not quite as much as our brain, but we don't trivially
mess with them.
But the excimer laser that made Lasik possible changed
that game tremendously.
Before the excemer laser, there were less than 20,000
eye surgeries done per year in the US.
Now we're over two million per year.
And the cost has gone down from 10s of thousands in the
distant past to a few hundred dollars per eye right now,
depending on whether or not you go with the very lowest
bidder on this.
Now is there an excimer laser for brain surgery?
I don't think there's anything quite that dramatic, but there
are technologies on the horizon that are dramatically
simplifying and speeding up brain surgery.
And the ones that I think about are neurovascular
stereotactic surgery.
So what does that mean?
Neurovascular means they don't cut open your skull, and
instead they get to the brain via the bloodstream.
So they insert something into normally the femoral artery,
and they can work this up.
This is actually a diagram for heart surgery.
They could work it all the way up into your brain.
And they've been doing this for aneurysms for the last
several years.
It has dramatically reduced the length of surgery, the
hospital stays, the cost, and the risk.
Because cutting open your skull is serious business.
The other part of this is what we call stereotactic surgery,
which means that the probe is being guided by magnets.
So it happens in a fancy-looking operating room
like this, probably not quite as clean and perfect,
hopefully with some doctors in it as well.
But the doctors actually sit at a terminal and use a
joystick to move the probe.
The probe itself has a camera on it so they can see.
There's an imager that shows them where it is on a
cross-section of the body.
And then the large device around this person is using a
shifting magnetic field to pull the probe along in the
direction the doctor wants to go with the joystick, rather
than this clunky old cutting and so on and so forth.
One of the real luminaries in neuroscience, a guy by the
name of Rodolfo Llinas, who's the editor of a journal called
Neuron, has proposed that we could use this to build a very
advanced brain-computer interface.
What he points out is that nanotubes are essentially
excellent wires, and are very, very, very small.
These aren't even nanotubes.
These are nanowires that are much bigger than a nanotube,
but they're still very, very, very small.
In fact, each of these is about a quarter of the
diameter of the wavelength of visible light.
Visible light is about 400-ish nanometers, and these are
about 50 nanometers across.
These are so small that a bundle of a million of them
would take up only about 1% of the cross sectional area of
the smallest capillary in your body.
So Llinas says, hey, let's stick in a bundle of a million
of these, work it up into your brain, and then use the
magnets to make it spread out and touch a million different
neurons in your brain all at once.
And he is starting to do animal studies investigating
this technology.
And we'll see if that's a viable path to dramatically
increase the bandwidth in and out of the brain.
And this gets closer to the technology of Nexis.
So those are the hardware issues.
But let's talk about software.
Some of us work on software, right?
We all know that software is absolutely perfect, especially
Microsoft software.
I'll point at myself a little bit here.
I've never seen this screen myself.
I don't know, it's a myth, I think.
But what does it mean if the software running in your
brain, or even just the software your brain is talking
to, suddenly crashes?
What does that mean if you get malware or a virus?
That could be a little bit more serious when it's
something running inside your brain.
And in fact some of this has been shown.
Anyone know who this is?
This is a fellow named Barnaby Jack.
He's famous mostly for this hack he did on the stage at
Black Hat I think in 2010, where from across the stage,
he made this ATM just start spewing out bills.
But lately Barnaby's been interested in medical devices
and the security around them.
So in 2011 he did this hack.
What he's got in his hand is a custom-built antenna,
essentially.
And what he's figured out that he can do is that wirelessly,
he can discover any insulin pump by Medtronic, the number
one manufacturer, within several blocks of himself, can
sniff out their IDs, get root access, and make them do
whatever he wants, including pump out all 300 units of
insulin that they have inside of them, which for a diabetic
would be lethal.
That was 2011.
In 2012, he did another hack with a leading pacemaker model
where he demonstrated that in a smaller range, within about
50 feet, he could cause it to drop a 730-volt discharge into
the person wearing it, which would also be lethal.
And in fact, do any of you watch Homeland?
I'm going to give a spoiler, so go la, la, la if you don't
want to hear this.
In a recent--
I don't watch the show, but I'm told that in a recent
episode, they assassinated the vice president by remotely
turning off his pacemaker.
And if you recall, *** Cheney, when he was vice
president, for quite a while, had no pulse.
He was in continuous cardiac failure and instead he had an
implanted pump that moved his blood around.
I don't know what APIs that had.
But science fiction sounds more and more like real life.
So security is a very big challenge.
That's one of the things that the book gets into a little
bit, is what happens when people can reroute your brain
or somebody else's.
But the biggest question I think of all of the stuff, as
it comes to pass, as I think it will, is what's the impact
on society?
So the printing press is very old information technology and
it's had marvelous effects.
Let's talk about two effects.
It supercharged innovation.
By allowing ideas to spread from person to person, it just
accelerated the rate of invention, the rate of
knowledge dissemination.
And it didn't start the Renaissance.
Some other things did.
But it really powered through, and the scientific
revolution, and so on.
It also was a great tool of anti-authoritarianism.
So it's famous, the first thing printed on the printing
press was the Gutenberg Bible.
Definitely a tool of authority, if you will.
One of the later things printed on it was Martin
Luther's list of complaints about the Catholic Church, his
pamphlets which he then nailed to the doors of churches and
led to the creation of the Protestant Church and the
erosion of power of the Catholic Church.
It was also used to print the Revolutionary pamphlets in the
late 1700s of the founding fathers of this country.
So overall, information technology has been a very
anti-authoritarian thing in this context.
But it wasn't always so.
The very first information technologies created by the
Sumerians and Egyptians were in the hands
of a very few people.
And they more allowed for the creation of empire and the
clamping down on the freedoms of the individual than the
other way around, again, because only a small minority
had access to and control over and the ability to use this
technology.
So today we have that conflict about
information tech currently.
During the Arab Spring, this picture made its way around, I
think taken from a place in Egypt.
Revolution tools, no longer the AK-47, no longer the
machete, now Twitter and Facebook.
And to a certain extent, this is true.
Wael Ghonim, who was one of the people behind the Arab
Spring, talks about how one of the things that started this
was a video taken of police beating up a student in
Alexandria, which was then posted to YouTube and then
made its way onto Facebook.
And the first Arab Spring protest--
in Egypt, anyway.
It was the second country.
But the first ones there were organized as a protest about
this kid's beating, and it was a Facebook event.
On the other hand, people will tell you that in Iran, the
authorities have used Twitter and Facebook to track down
people who are saying anti-authoritarian things.
So we still have this big question of to what extent and
in what situations will information technology be used
as a tool of liberation versus a tool of building
something like 1984.
And that's the question I think I try to most get that
in hopefully a fun way in the novel.
And I am an optimist.
I think that by and large, this sort of technology, a
more intimate means of communicating than ever
before, will be a hugely positive thing for society.
But we have to be on the lookout for and guard against
its possible abuses as well.
So that's it.
Thank you very much.
I think we have time for questions.
And there's some books outside that I'm
happy to sign for people.
AUDIENCE: I've read that one of the problems with the
electrode connections is that they degrade over time.
Has that gotten better in recent years?
RAMEZ NAAM: Yeah, so it's a very big question.
In some ways, we don't know.
So the brain doesn't like the intrusion of electrodes.
It treats them as a foreign object.
It tries to guard against them.
It builds scar tissue.
So there's two frontiers on that.
One is making electrodes smaller
and smaller and smaller.
So the newest technologies of nano probes are about 100
times smaller than the current ones.
The other is what they call bio-compatible materials that
won't break down.
And there's whole labs dedicated to just that.
We're still not there yet, partially because no one has
worn one of these things for a very, very long time.
So in some ways, we don't see what all the effects will be
after you've had one for 20 or 30 years.
AUDIENCE: Would you like to make some predictions for us?
RAMEZ NAAM: Oh my gosh, predictions.
I'll make a couple predictions.
By 10 years from now, the first people with bionic eyes
will be here.
In fact, it's probably closer than that.
You're looking at about five years for approval of the
retinal prosthesis.
And within 10 years, I'd say there will be at least 100,000
people that have them.
In the next 10 years, the FDA will start acquiring threat
modeling and security validation of the software and
protocols on your devices.
There will be in the next 10 years the growth of a kind of
DIY hacking movement that wants to hack their retinal
prostheses and their cochlear implants and so on.
Those are a few.
AUDIENCE: Have people done stuff outside the spectrum
that we normally see or hear?
RAMEZ NAAM: Not exactly, but it is possible to do sort of
translation or shifting.
So it's totally possible to take one sensory domain and
shift it to another.
So we could take infrared with a rental prosthesis, pick it
up, and compress it into a very bright red, for instance,
on the screen.
There's no obstacle to that, really, at all.
And you can see that.
There's a brand new retinal prosthesis that's being
experimented with that does
something very funky, actually.
So normally, the connections in the retina aren't numerous
enough to give you the ability to read, really.
So this does something else.
It picks up sight.
It spots Braille.
It sees Braille, and it massively zooms the Braille,
effectively, and feeds that in so that blind people can take
all of the books they already have and actually look at them
and learn to read that way.
So there's all sorts of domain shifting that
you can imagine doing.
AUDIENCE: One last question.
When do you think we'll beat the bandwidth of
our eyes and ears?
RAMEZ NAAM: That's a really good question.
When do we think we'll be the bandwidth of
our eyes and ears.
I think it'll be a while, to be honest.
As much as I write about this and so on, progress on the
electronics goes very, very fast, but you're always more
conservative when you're doing trials on
humans or even on animals.
So I think it'll be on the outside of that 10-year
spectrum, but probably not 20.
BRAD TEMPLETON: There's a friend of mine who made a
smartphone app that remaps the colors just
in the visual spectrum.
It doesn't go outside of the visual spectrum, but it remaps
them so that they match more the colorblind person's eyes.
So the colorblind people can look through this and they can
now tell the difference between the red and the green
in the way they can't tell with their own eyes.
They can just hold up their phone and say,
oh, now I see that.
That's kind of cool.
Any more questions?
Yeah,
AUDIENCE: So I have a question about these neural implants
and learning.
I think in the book people pick it up almost
instantaneously.
But what do you think about learning or the ability for
somebody who's in their 50s to learn--
just to pick these things up quickly.
Do you have to be young, old?
Is there any--
RAMEZ NAAM: It's a good question.
It probably will be the case that being
younger will be an advantage.
But we're also getting better on the algorithm side.
And we're in that feedback is vastly important.
So the first monkey experiments, the monkey
couldn't actually see the robot arm moving.
And it never occurred to them this was very important.
So this was actually a very famous monkey experiment where
the monkey sends its brain signals over the internet to
robot arms 600 miles away, which was a great gimmick, but
it couldn't see the thing moving.
So it turns out that in most of these systems, the learning
happens in both directions.
The system has to adapt.
It has to be calibrated, usually
intentionally, for the person.
When they put in the brain visual implant for Jens, they
spend time calibrating, like learning what each pixel is,
effectively, on the grid.
And they just--
it goes into the software.
But it's also the case that the brain
adapts to these things.
And it can adapt pretty darn quickly.
And now Johnny Ray, the paralyzed guy, it took them
about six months to train him to use his
one electrode system.
Now they can get people working pretty well in a
matter of several minutes.
It's fast enough-- they've done some studies where they
have a brain implant for people who are either testing
it or people who are having surgery for other reasons,
usually epileptics having oblation done.
They ask their permission to do this, and
the people sign off.
They'll put the thing on the surface of their brain during
the surgery, and they're have them playing Pong with it now
within 15 minutes before taking it off, in large part
because the algorithms that are interpreting what's
happening have gotten a lot better.
Which really means we've gotten better understanding of
what the encoding is of these data types, if you will,
inside the brain.
AUDIENCE: Cool.
RAMEZ NAAM: Yeah.
AUDIENCE: --books from, I don't know, the '70s, '80s?
He was writing about people who had brain implants that
stimulated their pleasure centers, since they turn into
wire heads and sat and basically had a *** high
continuously forever.
What do you see--
I mean obviously this could be applied to mood and a variety
of things other than just information.
What do you see happening there and
how would it be regulated?
RAMEZ NAAM: Absolutely.
So in one of the early experiments, those creepy
guys, Delgado and Heath, they did some
very, very creepy things.
One of the things they did was
simulation of pleasure centers.
They used it in a couple ways.
They tried to switch a guy from gay to straight.
By in a psych ward, they put something in
his pleasure center.
They brought in a female *** and then repeatedly
stimulated his pleasure center.
It didn't work, but they were thinking that way.
They gave another guy the control over his pleasure
center, and found at some times he would hit the button
1,500 times an hour, which means about
once every two seconds.
But the ultimate control is--
AUDIENCE: [INAUDIBLE] slow or something?
RAMEZ NAAM: I think maybe he got really blissed out for a
second and then he'd come to and be like, oh yeah.
So the ultimate control is, don't do the surgery to
implant the electrode there.
And I think that that will be the medical ethical norm in
most of these cases.
But there are certainly potentials for abuse.
And actually in the sequel to Nexus, I show a character that
has a very serious problem that's developed as a result
of maybe not having the implants avoid certain areas.
AUDIENCE: I think--
I heard a technology called transcranial magnetic
stimulation.
So that seems a little bit safer, maybe, because it's
completely external.
So do you have any--
RAMEZ NAAM: With transcranial magnetic stimulation, and now
with direct current cranial stimulation, they've done some
interesting things.
With the magnetic stuff, a lot of it has been--
well one group can make at least some people have a
religious experience or something like that by
stimulating the right areas.
And then with the electrical stuff, they can also,
apparently, at least they're claiming make people somewhat
autistic and have almost savant-like
abilities appear in some.
It's kind of very early days.
You're not sure how much you trust the research.
You want to see 5 or 10 papers on something first.
With the direct current simulation, something that
does look fairly repeatable is that they can
speed learning time.
So they can maybe cut your learning time on certain
things by 20% by running a very, very mild current across
the top of your skull while you're learning something.
So that looks pretty neat.
But it's not a technology that's going to get you to
have, like video in or video out or something like that.
BRAD TEMPLETON: The direct current simulation actually
makes you more trusting, too.
You're more--
so I wanted, when I go into negotiations now, I say, would
you mind wearing this hat while we have this
negotiation?
Nobody wants to wear the hat.
Are we mostly done with audience questions?
Matt has a question.
RAMEZ NAAM: Oh.
AUDIENCE: So what do we know about the encoding?
With pixels, it's a linear kind of thing.
In the brain, is it logarithmic?
Is it linear?
How does that all?
RAMEZ NAAM: So it differs by sense.
And the senses are going to be a lot easier than the higher
functions because we can kind of understand
what the coding means.
So in vision it is pretty much spatial, in V1, the visual
area in the back of your head.
But it's spatial with some caveats.
Things are not entirely spatial.
It's kind of more spatial if you took the grid and
flattened it.
But then your arms, of course, are all kind of
squiggly and so on.
And then some neurons, they are more including certain
colors, and more encoding motion, so with those caveats.
And we kind of--
I wouldn't say there's a formalism of that encoding.
It's just that we've gotten better and better at having
code that can actually translate that.
In audio, we think that the encoding is
the frequency domain.
So as you shift from one region to another, it gets
more and more towards representation of higher and
higher frequencies.
In motion, it's more complex than that.
But one thing that's nice in all of these, like a general
characteristic we're finding is that we can get good
results with very sparse connections.
In any given region, we're sampling a
very, very tiny bit.
So it's not a very fragile protocol.
It's a very robust protocol that just gets higher and
higher resolution as you have access to more of
what's going on.
And there's other stuff we know in the higher domains
even, we know some various different things.
Like in parts of your brain, you can trace the recognition
of different animals through one
particular area of the brain.
And you can find very reliably that the cat versus dog areas
are kind of separate and in the same relationship
physically in different people's brains.
And there's even weird studies with taking monkeys, having
them look at pictures of a cat, stimulating the dog
center, and getting them to recognize it as a dog instead.
So there is very, very, very funky stuff out there.
BRAD TEMPLETON: So there's a physical difference between
cat people and dog people then.
RAMEZ NAAM: I've never seen a correlation of is that area
enlarged in one or the other, but that would be a great
study, Brad.
Let's put in a grant.
BRAD TEMPLETON: That's right.
I want to ask about mind control, because you love mind
control and it plays a role in the book.
And not just, though, in forced mind control, but the
concept of voluntary mind control, people wanting to
change their attitudes, people wanting to--
or socially pressured to do so.
For example, there's a hormone that makes some animals more
monogamous.
If that applied-- or a gene therapy like that applied in
humans, you could see it being part of the wedding vows.
Why won't you take this treatment that makes you love
me forever, darling?
You love me, right?
What do you think about the potential for mind control
from these?
RAMEZ NAAM: That's the vasopressin
receptor gene, right?
Naked voles are one of the only annals we found that's
really monogamous.
Yeah, I think people will do it.
There was a time in my life that I would have-- that I
felt like I was not very outgoing or not very
adventurous.
And if I had an option to change my dopamine DDR2
receptor and suddenly want to go skydiving all the time, I
probably would have done it.
I might now.
So I think people will voluntarily engage in
technologies that will allow them to change their own
personalities all the time.
Or for that matter, I want something that's going to get
me out of bed and to the gym reliably every day.
It's better than a workout buddy.
You've got a little app on your brain that's like, up,
here we go.
AUDIENCE: So question on that vein.
It's already bewildering difficult to keep track of the
options we have in life, with the internet and so forth.
So this is a future where everybody has to understand
how the brain works and what their options are
for changing it.
RAMEZ NAAM: Not necessarily.
I would say most people who use apps and their phone don't
understand how their phone works.
They have to have some understanding of some
abstractions about the phone.
But they don't have to be neuroscientists of the phone.
They don't have to be EEs to know, like, oh, I
really like this app.
Or no, this kind of thing is possible, they have to figure
out some things.
Like, eventually people really need to understand, is the
data in the cloud or on their machine?
Because someday they might not have signal.
So they have to learn some things about what's going on.
But they don't have to learn all the deep details.
But I agree that the future gets more and more bewildering
and that's a challenge.
AUDIENCE: I wonder if we know anything about memory.
Like is there some way--
is any research into, like, adding a hard drive to your
brain, basically?
Because you've talked a lot about processors.
RAMEZ NAAM: That's what people want to do.
I think it's going to be tough.
We don't understand a whole lot.
But just like with vision and sound and motion, the way that
we've actually made progress on the encoding is trying to
repair these things in people that have some sort of damage.
So that's the most encouraging thing about the rat study and
so on, is that doing the work to try to fix the damage will
functionally teach us more and more about the encoding.
But today I'd say we know relatively
little, to be honest.
BRAD TEMPLETON: We used to know more yesterday.
RAMEZ NAAM: It's possible, some of us.
AUDIENCE: How does this change your ideas about death?
And what do you--
let's say we build AI machines in the future.
How do we deal with the problem that our biological
bodies are going to die?
RAMEZ NAAM: Well it's not really exactly in this domain.
But I guess it's a large extension of it, is I believe
uploading is possible.
Everything we know says that if you had a complete map of
your brain down to the molecular level, at least, and
we could simulate it, we'd have a copy of
your thoughts running.
We don't know how technically difficult it is.
You can make a guess.
You can say, oh, if we get it at the level of neurons and
synapses, that's probably good enough.
We don't know.
We won't know until we get there and try it, to be
totally honest.
So I think eventually technology
will make that possible.
Ray Kurzweil thinks it's going to be 30 years from now.
It might be 100 years from now.
It might be 1,000 years now.
But the brain is just another information processor, a very
messy one, a very interesting one.
And it's really, the pattern of information is you, to a
certain extent.
So we can eventually decouple those.
Just exactly when is still a very open questions.
AUDIENCE: Would you feel like that was you?
RAMEZ NAAM: That copy of me would feel that it was me.
That copy of me would remember it being me right here
answering this question, and be like, oh, I feel like me.
And the other copy of me would eventually be dead.
So now, what if that copy made 1,000 more copies?
Would they all feel like me?
Yes.
We will have to-- here's the philosophical thing I'll say.
Right now, if I'll use computer lingo a little bit,
we don't differentiate, when we talk about me, between the
class and the instance of the class.
So eventually our conception of personhood will change.
And we'll see, like, do you believe that all other objects
that are of your class are the same as you?
In some ways, yes.
And we'll start to make that differentiation.
But we're not ready for that yet.
Most philosophers don't have an object-oriented programming
background, so we're not quite ready for that.
BRAD TEMPLETON: Well thanks a lot, Mez, for coming in.
RAMEZ NAAM: Awesome.
Thank you all.