This essay, originally published in
eight short parts, aims to condense the current knowledge on
It explores the state of AI development,
overviews its challenges and dangers, features work by the most
significant scientists, and describes the main predictions of
possible AI outcomes. This project is an adaptation and major
shortening of the two-part essay
AI Revolution by Tim Urban of Wait But Why.
I shortened it by a
factor of 3, recreated all images, and tweaked it a bit.
Assuming that human scientific activity
continues without major disruptions, artificial intelligence may
become either the most positive transformation of our history or, as
many fear, our most dangerous invention of all.
AI research is on a
steady path to develop a computer that has cognitive abilities equal
to the human brain, most likely within three decades (timeline in
From what most AI scientists predict, this invention may
enable very rapid improvements (called
fast take-off), toward
something much more powerful - Artificial Super Intelligence
entity smarter than all of humanity combined (more on ASI in chapter
We are not talking about some imaginary future. The first level
of AI development is gradually appearing in the technology we use
everyday. With every coming
year these advancements will accelerate and the technology will
become more complex, addictive, and ubiquitous.
We will continue to
outsource more and more kinds of mental work to computers,
disrupting every part of our reality: the way we organize ourselves
and our work, form communities, and experience the world.
The Guiding Principle Behind
To more intuitively grasp the guiding
principles of AI revolution, let's first step away from scientific
Let me invite you to take part in a story.
you've received a time machine and been given a quest to bring
somebody from the past. The goal is to shock them by showing them
the technological and cultural advancements of our time, to such a
degree that this person would perform SAFD (Spinning Around From
So you wonder which era should you
time-travel to, and decide to hop back around 200 years.
You get to
the early 1800s, retrieve a guy and bring him back to 2016.
"…walk him around and watch him react to everything. It's impossible
for us to understand what it would be like for him to see shiny
capsules racing by on a highway, talk to people who had been on the
other side of the ocean earlier in the day, watch sports that were
being played 1,000 miles away, hear a musical performance that
happened 50 years ago, and play with …[a] magical wizard rectangle
that he could use to capture a real-life image or record a living
moment, generate a map with a paranormal moving blue dot that shows
him where he is, look at someone's face and chat with them even
though they're on the other side of the country, and worlds of other
inconceivable sorcery." 1
It doesn't take much. After two minutes he is SAFDing.
Now, both of you want to try the same
thing, see somebody Spinning Around From Disbelief, but in your new
friend's era. Since 200 years worked, you jump back to the 1600s and
bring a guy to the 1800s.
He's certainly genuinely interested in
what he sees. However, you feel it with confidence - SAFD will never
happen to him. You feel that you need to jump back again, but
somewhere radically further.
You settle on rewinding the clock
15,000 years, to the times,
"…before the First Agricultural
Revolution gave rise to the first cities and the concept of
You bring someone from the hunter-gatherer world and show him,
vast human empires of 1750 with their towering churches, their
ocean-crossing ships, their concept of being 'inside,' and their
enormous mountain of collective, accumulated human knowledge and
discovery" 3 - in
forms of books. It doesn't take much.
He is SAFDing in the first two
Now there are three of you, enormously
excited to do it again.
You know that it doesn't make sense to go
back another 15,000, 30,000 or 45,000 years. You have to jump back,
again, radically further. So you pick up a guy from 100,000 years
ago and you walk with him into large tribes with organized,
sophisticated social hierarchies.
He encounters a variety of hunting
weapons, sophisticated tools, sees fire and for the first time
experiences language in the form of signs and sounds. You get the
idea, it has to be immensely mind-blowing.
He is SAFDing after two
So what happened? Why did the last guy
had to hop → 100,000 years, the next one → 15,000 years, and the guy
who was hopping to our times only → 200 years?
"This happens because more
advanced societies have the ability to progress at a faster rate
than less advanced societies - because they're more advanced.
[1800s] humanity knew more and had better technology…"
...so it's no wonder they could make further advancements than humanity
from 15,000 years ago.
The time to achieve SAFD shrank from ~100,000
years to ~200 years and if we look into the future it will rapidly
shrink even further.
Ray Kurzweil, AI expert and scientist, predicts
"…20th century's worth of progress happened between 2000 and
2014 and that another 20th century's worth of progress will happen
by 2021, in only seven years 5…
couple decades later, he believes a 20th century's worth of progress
will happen multiple times in the same year, and even later, in less
than one month 6…Kurzweil
believes that the 21st century will achieve 1,000 times the progress
of the 20th century."
"Logic also suggests that if the
most advanced species on a planet keeps making larger and larger
leaps forward at an ever-faster rate, at some point, they'll make a
leap so great that it completely alters life as they know it and the
perception they have of what it means to be a human.
Kind of like
how evolution kept making great leaps toward intelligence until
finally it made such a large leap to the human being that it
completely altered what it meant for any creature to live on planet
And if you spend some time reading about what's going
on today in science and technology, you start to see a lot of signs
quietly hinting that life as we currently know it cannot withstand
the leap that's coming next."8
The Road to Artificial General
Building a Computer as Smart as Humans
Artificial Intelligence, or AI, is a
broad term for the advancement of intelligence in computers.
varied opinions on this topic, most experts agree that there are
three categories, or calibers, of AI development.
ANI: Artificial Narrow Intelligence
1st intelligence caliber. "AI that
specializes in one area. There's AI that can beat the world chess
champion in chess, but that's the only thing it does."
AGI: Artificial General Intelligence
2nd intelligence caliber. AI that
reaches and then passes the intelligence level of a human, meaning
it has the ability to "reason, plan, solve problems, think
abstractly, comprehend complex ideas, learn quickly, and learn from
ASI: Artificial Super Intelligence
3rd intelligence caliber. AI that
achieves a level of intelligence smarter than all of humanity
combined - "ranging from just a little smarter... to one trillion
times smarter." 11
Where are we currently?
"As of now, humans have conquered
the lowest caliber of AI - ANI - in many ways, and it's everywhere:"
"Cars are full of ANI systems,
from the computer that figures out when the anti-lock brakes
kick in, to the computer that tunes the parameters of the fuel
"Google search is one large
ANI brain with incredibly sophisticated methods for ranking
pages and figuring out what to show you in particular. Same goes
for Facebook's Newsfeed."
Email spam filters "start off
loaded with intelligence about how to figure out what's spam and
what's not, and then it learns and tailors its intelligence to
your particular preferences."
Passenger planes are flown almost
entirely by ANI, without the help of humans.
self-driving car, which is being tested now, will contain
robust ANI systems that allow it to perceive and react to the
world around it."
"Your phone is a little ANI
factory... you navigate using your map app, receive tailored
music recommendations from Pandora, check tomorrow's weather,
talk to Siri." 17
"The world's best Checkers,
Chess, Scrabble, Backgammon, and Othello players are now all ANI
"Sophisticated ANI systems are
widely used in sectors and industries like military,
manufacturing, and finance (algorithmic high-frequency AI
traders account for more than half of equity shares traded on US
ANI systems as they are now
aren't especially scary.
At worst, a glitchy or badly-programmed ANI
can cause an isolated catastrophe like"
a plane crash, a nuclear power plant malfunction, or,
markets disaster (like the
2010 Flash Crash when an ANI program reacted the wrong way to an
unexpected situation and caused the stock market to briefly plummet,
taking $1 trillion of market value with it, only part of which was
recovered when the mistake was corrected)...
But while ANI doesn't
have the capability to cause an existential threat, we should see
this increasingly large and complex ecosystem of relatively-harmless ANI as a precursor of the world-altering hurricane that's on the
Each new ANI innovation quietly adds another brick onto the
road to AGI and ASI."
What's Next? Challenges Behind
"Nothing will make you
appreciate human intelligence like learning about how
unbelievably challenging it is to try to create a computer as
smart as we are…
Build a computer that can multiply ten-digit
numbers in a split second - incredibly easy.
Build one that can
look at a dog and answer whether it's a dog or a
cat - spectacularly difficult. Make AI that can beat any human
in chess? Done. Make one that can read a paragraph from a
six-year-old's picture book and not just recognize the words but
understand the meaning of them?
Google is currently spending
billions of dollars trying to do it."
Why are "hard things - like
calculus, financial market strategy, and language translation... mind-numbingly easy for a computer, while easy things - like
vision, motion, movement, and perception - are insanely hard for
"Things that seem easy to us
are actually unbelievably complicated.
They only seem easy
because those skills have been optimized in us (and most
animals) by hundreds of million years of animal evolution.
When you reach your hand up
toward an object, the muscles, tendons, and bones in your
shoulder, elbow, and wrist instantly perform a long series
of physics operations, in conjunction with your eyes, to
allow you to move your hand in a straight line through three
On the other hand, multiplying big numbers or
playing chess are new activities for biological creatures and we
haven't had any time to evolve a proficiency at them, so a
computer doesn't need to work too hard to beat us."
One fun example…
When you look at picture A,
and a computer both can figure out that it's a rectangle with two
distinct shades, alternating. Tied so far."
"You have no problem
giving a full description of the various opaque and translucent
cylinders, slats, and 3-D corners, but the computer would fail
miserably. It would describe what it sees - a variety of
two-dimensional shapes in several different shades - which is
actually what's there." 27
"Your brain is doing a ton of fancy shit to interpret the implied
depth, shade-mixing, and room lighting the picture is trying to
Looking at picture C,
"a computer sees a
two-dimensional white, black, and gray collage, while you easily see
what it really is" 29 - a
photo of a girl and a dog standing on a rocky shore.
"And everything we just mentioned is
still only taking in visual information and processing it. To be
human-level intelligent, a computer would have to understand things
like the difference between subtle facial expressions, the
distinction between being pleased, relieved and content" 30.
How will computers reach even higher
abilities like complex reasoning, interpreting data, and associating
ideas from separate fields (domain-general
"Building skyscrapers, putting
humans in space, figuring out the details of how the Big Bang went
down - all far easier than understanding our own brain or how to
make something as cool as it.
As of now, the human brain is the most
complex object in the known universe."
If an artificial intelligence is
going to be as intelligent as the human brain, one crucial thing has
to happen - the AI,
"needs to equal the brain's raw computing
capacity. One way to express this capacity is in the total
calculations per second the brain could manage."
The challenge is that currently
only a few of the brain's regions are precisely measured.
Ray Kurzweil, has developed a method for estimating the total cps of
the human brain.
He arrived at this estimate by taking the cps from
one brain region and multiplying it proportionally to the weight of
that region, compared to the weight of the whole brain.
"He did this
a bunch of times with various professional estimates of different
regions, and the total always arrived in the same ballpark - around
1016, or 10 quadrillion cps."
"Currently, the world's fastest
Tianhe-2, has actually beaten that number, clocking in at about
34 quadrillion cps."
"taking up 720 square meters of
space, using 24 megawatts of power (the brain runs on just
20 watts), and costing $390 million to build. Not especially
applicable to wide usage, or even most commercial or industrial
usage yet." 35
"Kurzweil suggests that we think
about the state of computers by looking at how many cps you can buy
for $1,000. When that number reaches human-level - 10 quadrillion
cps - then that'll mean AGI could become a very real part of life."
Currently we're only at about 1010
(10 trillion) cps per $1,000.
"that the world's maximum computing power
doubles approximately every two years, meaning computer hardware
advancement, like general human advancement through history, grows
right on pace with this graph's predicted trajectory:"
"puts us right on
pace to get to an affordable computer by 2025 that rivals the power
of the brain...
But raw computational power alone doesn't make a
computer generally intelligent - the next question is, how do we
bring human-level intelligence to all that power?"
The hardest part of creating AGI is
learning how to develop its software.
"The truth is, no one really
knows how to make it smart - we're still debating how to make a
computer human-level intelligent and capable of knowing what a dog
and a weird-written B and a mediocre movie is." 40
But there are a couple of strategies.
These are the three most
1. Copy how the brain works
The most straight-forward idea is
to plagiarize the brain, and build the computer's architecture with
close resemblance to how a brain is structured.
artificial neural network. It starts out as a network of transistor
'neurons,' connected to each other with inputs and outputs, and it
knows nothing - like an infant brain.
The way it 'learns' is it
tries to do a task, say handwriting recognition, and at first, its
neural firings and subsequent guesses at deciphering each letter
will be completely random.
But when it's told it got something
right, the transistor connections in the firing pathways that
happened to create that answer are strengthened; when it's told it
was wrong, those pathways' connections are weakened.
After a lot of
this trial and feedback, the network has, by itself, formed smart
neural pathways and the machine has become optimized for the task."
The second, more radical approach to
plagiarism is whole brain
Scientists take a real brain, cut it into a large
number of tiny slices to look at the neural connections and
replicate them in a computer as software. If that method is ever
successful, we will have,
"a computer officially capable of
everything the brain is capable of - it would just need to learn and
How far are we from achieving whole brain
emulation? Well so far, we've
just recently been able to emulate a 1mm-long flatworm brain,
which consists of just 302 total neurons." 42
To put this into perspective, the human brain consists of
86 billion neurons linked by trillions of synapses.
2. Introduce evolution to computers
"The fact is, even if we can
emulate a brain, that might be like trying to build an airplane by
copying a bird's wing-flapping motions - often, machines are best
designed using a fresh, machine-oriented approach, not by mimicking
If the brain is just too complex for us to digitally replicate, we
could try to emulate evolution instead.
This uses a process called
group of computers would try to do tasks, and the most successful
ones would be bred with
each other by having half of each of their programming merged
together into a new computer.
The less successful ones would be
Speed and a goal-oriented approach are the advantages that
artificial evolution has over biological evolution.
"Over many, many
iterations, this natural selection process would produce better and
The challenge would be creating an automated
evaluation and breeding cycle so this evolution process could run on
its own." 45
3. "Make this whole thing the
computer's problem, not ours"
The last concept is the simplest, but
probably the scariest of them all.
"We'd build a computer whose two
major skills would be doing research on AI and coding changes into
itself - allowing it to not only learn but to improve its own
We'd teach computers to be computer scientists so they
could bootstrap their own development." 47
This is the likeliest way to get AGI soon that we know of.
All these software advances may
seem slow or a little bit intangible, but as it is with the
sciences, one minor innovation can suddenly accelerate the pace of
Kind of like the aftermath of the Copernican
revolution - the discovery that suddenly made all the complicated
mathematics of the planets' trajectories much easier to calculate,
which enabled a multitude of other innovations.
"exponential growth is intense and what seems like a snail's pace of
advancement can quickly race upwards."
It's very real that at some point we
will achieve AGI:
software that has achieved human-level, or
beyond human-level, intelligence.
Does this mean that at that
very moment the computers will be equally capable as us?
Actually, not at all - computers will be way more efficient.
Because of the fact that they are electronic, they will have
neurons max out at around 200 Hz, while today's
microprocessors... run at 2 GHz, or 10 million times faster."
Forgetting or confusing
things is much harder in an artificial world. Computers can
memorize more things in one second than a human can in ten
years. A computer's memory is also more precise and has a
much greater storage capacity.
transistors are more accurate than biological neurons, and
they're less likely to deteriorate (and can be repaired or
replaced if they do). Human brains also get fatigued easily,
while computers can run nonstop, at peak performance, 24/7."
work is ridiculously challenging because of time-consuming
communication and complex social hierarchies. The bigger the
group gets, the slower the output of each person becomes.
AI, on the other hand, isn't biologically constrained to one
body, won't have human cooperation problems, and is able to
synchronize and update its own operating system.
We need to realize that AI,
"wouldn't see 'human-level intelligence' as some important
milestone - it's only a relevant marker from our point of
view - and wouldn't have any reason to 'stop' at our level.
given the advantages over us that even human
intelligence-equivalent AGI would have, it's pretty obvious that
it would only hit human intelligence for a brief instant before
racing onwards to the realm of superior-to-human intelligence."
The true distinction between
humans and ASI wouldn't be its advantage in intelligence speed,
"in intelligence quality
- which is something completely
What makes humans so much more intellectually capable
than chimps isn't a difference in thinking speed - it's that
human brains contain a number of sophisticated cognitive modules
that enable things like complex linguistic representations or long-term planning or abstract reasoning, that chimps' brains do
Speeding up a chimp's brain by thousands of times
wouldn't bring him to our level - even with a decade's time of
learning, he wouldn't be able to figure out how to... "
assemble a semi-complicated Lego model by looking at its
manual - something a young human could achieve in a few minutes.
"There are worlds of human cognitive function a chimp will
simply never be capable of, no matter how much time he spends
"And in the scheme of the
biological intelligence range... the chimp-to-human quality
intelligence gap is tiny."
In order to render how big a deal
it would be to exist with something that has a higher quality of
intelligence than us, we need to imagine AI on the intelligence
staircase two steps above us:
"its increased cognitive ability over
us would be as vast as the chimp-human gap... And like the chimp's
incapacity to ever absorb …"
what kind of magic happens in the mechanism of a doorknob - "we will
never be able to even comprehend the things... [a machine of that
intelligence] can do, even if the machine tried to explain them to
us... And that's only two steps above us."
"A machine on the second-to-highest step
on that staircase would be to us as we are to ants." 59
"Superintelligence of that magnitude is not something we can
remotely grasp, any more than a bumblebee can wrap its head around
Keynesian Economics. In our world, smart means a 130 IQ and stupid
means an 85 IQ - we don't have a word for an IQ of 12,952."
"But the kind of superintelligence
we're talking about today is something far beyond anything on this
staircase. In an intelligence explosion - where the smarter a
machine gets, the quicker it's able to increase its own
intelligence - a machine might take years to rise from...
intelligence of an ant to the intelligence of the average human, but
it might take only another 40 days to become Einstein-smart.
"it works to improve its intelligence, with an
Einstein-level intellect, it has an easier time and can make bigger
leaps. These leaps will make it much smarter than any human,
allowing it to make even bigger leaps."
From then on, following the rule of
exponential advancements and utilizing the speed and efficiency of
electrical circuits, it may perhaps take only 20 minutes to jump
"and by the time it's ten steps above us, it might be
jumping up in four-step leaps every second that goes by.
why we need to realize that it's distinctly possible that very
shortly after the big news story about the first machine reaching
human-level AGI, we might be facing the reality of coexisting on the
Earth with something that's here on the staircase (or maybe a
million times higher):" 63
"And since we just established
that it's a hopeless activity to try to understand the power of a
machine only two steps above us, let's very concretely state once
and for all that there is no way to know what ASI will do or what
the consequences will be for us.
Anyone who pretends otherwise
doesn't understand what superintelligence means."
"If our meager brains were able to
invent Wi-Fi, then something 100 or 1,000 or 1 billion times smarter
than we are should have no problem controlling the positioning of
each and every atom in the world in any way it likes, at any
time - everything we consider magic, every power we imagine a
supreme God to have will be as mundane an activity for the ASI as
flipping on a light switch is for us."
"As far as we're concerned, if an
ASI comes into being, there is now an omnipotent God on Earth
the all-important question for us is:
Will it be a good god?"
Let's start from the brighter side of
How Can ASI Change our World?
Speculations on Two Revolutionary
Nanotechnology is an idea that comes up
"in almost everything you read about the future of AI."
It's the technology that works at the nano scale - from 1 to 100
"A nanometer is a millionth of a millimeter. 1 nm-100 nm
range encompasses viruses (100 nm across), DNA (10 nm wide), and
things as small as molecules like hemoglobin (5 nm) and medium
molecules like glucose (1 nm).
If/when we conquer nanotechnology,
the next step will be the ability to manipulate individual atoms,
which are only one order of magnitude smaller (~.1 nm)." 68
To put this into perspective, imagine a
very tall human standing on the earth, with a head that reaches the
International Space Station (431 km/268 mi high).
The giant is
reaching down with his hand (30 km/19 mi across) to build,
using materials between the size of a grain of sand [.25 mm] and an
eyeball [2.5 cm]." 69
"Once we get nanotechnology
down, we can use it to make tech devices, clothing, food, a
variety of bio-related products - artificial blood cells, tiny
virus or cancer-cell destroyers, muscle tissue, etc. - anything
And in a world that uses nanotechnology, the cost of a
material is no longer tied to its scarcity or the difficulty of
its manufacturing process, but instead determined by how
complicated its atomic structure is.
In a nanotech world, a
diamond might be cheaper than a pencil eraser."
One of the proposed methods of
nanotech assembly is to make,
"one that could self-replicate, and
then let the reproduction process turn that one into two, those
two then turn into four, four into eight, and in about a day,
there'd be a few trillion of them ready to go."
But what if this process goes
wrong or terrorists manage to get a hold of the technology?
imagine a scenario where nanobots,
"would be designed to consume
any carbon-based material in order to feed the replication
process, and unpleasantly, all life is carbon-based. The Earth's
biomass contains about 1045 carbon atoms.
A nanobot would
consist of about 106 carbon atoms, so it would
take 1039 nanobots to consume all life on Earth, which would happen in 130
Scientists think a nanobot could replicate in
about 100 seconds, meaning this simple mistake would
inconveniently end all life on Earth in 3.5 hours."
We are not yet capable of
harnessing nanotechnology - for good or for bad.
"And it's not
clear if we're underestimating, or overestimating, how hard it
will be to get there. But we don't seem to be that far away. Kurzweil predicts that we'll get there by the 2020s.
know that nanotech could be an Earth-shaking development... The
US, the EU, and Japan 74
have invested over a combined $5 billion so far"
"Because everyone has always died,
we live under the assumption... that death is inevitable. We
think of aging like time - both keep moving and there's nothing
you can do to stop it."
For centuries, poets and philosophers have wondered if
consciousness doesn't have to go the way of the body.
describes us as,
"a soul fastened to a dying animal."
Richard Feynman, Nobel awarded physicists, views death from a
purely scientific standpoint:
"It is one of the most remarkable things that in all of the
biological sciences there is no clue as to the necessity of
If you say we want to make perpetual motion, we have
discovered enough laws as we studied physics to see that it
is either absolutely impossible or else the laws are wrong.
But there is nothing in biology yet found that indicates the
inevitability of death.
This suggests to me that it is not
at all inevitable, and that it is only a matter of time
before the biologists discover what it is that is causing us
the trouble and that that terrible universal disease or
temporariness of the human's body will be cured."
Theory of great species attractors
When we look at the history of
biological life on earth, so far 99.9% of species have gone
Nick Bostrom, Oxford professor and AI specialist,
"calls extinction an attractor state - a place species are... falling into and from which no species ever returns."
"And while most AI scientists... acknowledge that ASI would have the ability to send humans to
extinction, many also believe that if used beneficially, ASI's
abilities could be used to bring individual humans, and the species
as a whole, to a second
attractor state - species immortality."
"Evolution had no good reason to
extend our life-spans any longer than they are now... From an
evolutionary point of view, the whole human species can thrive
with a 30+ year lifespan" for each single human.
enough to reproduce and raise children... so there's no reason
for mutations toward unusually long life being favored in the
natural selection process."
"if you perfectly repaired or replaced a car's parts whenever
one of them began to wear down, the car would run forever. The
human body isn't any different - just far more complex...
seems absurd - but the body is just a bunch of atoms…"
...making up organically programmed DNA, which it is theoretically
possible to manipulate.
And something as powerful as ASI could
help us master genetic engineering.
Ray Kurzweil believes that,
"artificial materials will be integrated into the body more and
more... Organs could be replaced by super-advanced machine
versions that would run forever and never fail."
Red blood cells could be perfected by,
"red blood cell nanobots,
who could power their own movement, eliminating the need for a
heart at all...
Nanotech theorist Robert A. Freitas has already
designed blood cell replacements that, if one day implemented in
the body, would allow a human to sprint for 15 minutes without
taking a breath...
[Kurzweil] even gets to the brain and believes
enhance our mental activities to the point where humans will
be able to think billions of times faster"
by integrating electrical components and being able to access
"Eventually, Kurzweil believes
humans will reach a point when they're entirely artificial, a
time when we'll look back at biological material and think how
it was that humans were ever made of that
that humans aged, suffered from cancer, allowed random factors
like microbes, diseases, accidents to harm us or make us
When Will The First Machine Become
Predictions from Top AI Experts
How long until the first machine
reaches superintelligence? Not shockingly, opinions vary wildly, and
this is a heated debate among scientists and thinkers.
"Those people subscribe to the
belief that this is happening
soon - that
exponential growth is at work and machine learning, though
only slowly creeping up on us now, will blow right past us
within the next few decades.
"The Kurzweil camp would
counter that the only underestimating that's happening
is the under-appreciation of exponential growth, and they'd
compare the doubters to those who looked at the slow-growing
seedling of the internet in 1985 and argued that there was
no way it would amount to anything impactful in the near
"The doubters might argue back
that the progress needed to make advancements in
intelligence also grows exponentially harder with each
subsequent step, which will cancel out the typical
exponential nature of technological progress. And so on.
"A third camp, which includes
Nick Bostrom, believes neither group has any ground to
feel certain about the timeline and acknowledges both A)
that this could absolutely happen in the near future and B)
that there's no guarantee about that; it could also take a
much longer time.
"Still others, like philosopher
Hubert Dreyfus, believe all three of these groups are
naive for believing that there is potential of ASI, arguing
that it's more likely that it won't actually ever be
"So what do you get when
you put all of these opinions together?"
Timeline for Artificial General
"In 2013, Vincent C.
Müller and Nick Bostrom conducted a survey that asked
hundreds of AI experts... the following:"
"For the purposes of this question, assume
that human scientific activity continues without major
negative disruption. By what year would you see a (10% / 50%
/ 90%) probability for such Human-Level Machine Intelligence
[or what we call AGI] to exist?"
"asked them to name
an optimistic year (one in which they believe there's a 10%
chance we'll have AGI), a realistic guess (a year they
believe there's a 50% chance of AGI - i.e. after that year
they think it's more likely than not that we'll have AGI),
and a safe guess (the earliest year by which they can say
with 90% certainty we'll have AGI).
Gathered together as one
data set, here were the results:
Median optimistic year (10%
likelihood) → 2022 Median realistic year (50% likelihood) → 2040 Median pessimistic year (90% likelihood) → 2075
"So the median participant
thinks it's more likely than not that we'll have AGI 25
years from now.
The 90% median answer of 2075 means that if
you're a teenager right now, the median respondent, along
with over half of the group of AI experts, is almost certain AGI will happen within your lifetime.
"A separate study,
conducted recently by author James Barrat at Ben Goertzel's
annual AGI Conference, did away with percentages and simply
asked when participants thought AGI would be achieved
2030, by 2050, by 2100, after 2100, or never.
42% of respondents → By 2030 25% of respondents → By 2050 20% of respondents → By 2100 10% of respondents → After 2100 2% of respondents → Never
"Pretty similar to Müller
and Bostrom's outcomes. In Barrat's survey, over two thirds
of participants believe AGI will be here by 2050 and a
little less than half predict AGI within the next 15 years.
Also striking is that only 2% of those surveyed don't think AGI is part of our future."
Timeline for Artificial Super
"Müller and Bostrom also asked
the experts how likely they think it is that we'll reach ASI:
A), within two years of reaching AGI (i.e. an almost-immediate
intelligence explosion), and B), within 30 years."
Respondents were asked to choose a probability for each option.
Here are the results:
AGI-ASI transition in 2 years →
10% likelihood AGI-ASI transition in 30 years → 75% likelihood
"The median answer put a rapid (2
year) AGI-ASI transition at only a 10% likelihood, but a longer
transition of 30 years or less at a 75% likelihood.
know from this data the length of this transition [AGI-ASI] the
median participant would have put at a 50% likelihood, but for
ballpark purposes, based on the two answers above, let's
estimate that they'd have said 20 years.
"So the median opinion - the one
right in the center of the world of AI experts - believes the
most realistic guess for when we'll hit ASI... is [the 2040
prediction for AGI + our estimated prediction of a 20-year
transition from AGI to ASI] = 2060.
"Of course, all of the above
statistics are speculative, and they're only representative of the
median opinion of the AI expert community, but it tells us that a
large portion of the people who know the most about this topic would
agree that 2060 is a very reasonable estimate for the arrival of
potentially world-altering ASI.
Only 45 years from now"
Two Main Groups of AI Scientists with
Two Radically Opposed Conclusions
1 - The Confident Corner
Most of what we have discussed so far
represents a surprisingly large group of scientists that share
optimistic views on the outcome of AI development.
confidence comes from is up for debate. Critics believe it comes
from an excitement so blinding that they simply ignore or deny
potential negative outcomes.
But the believers say it's naive to
conjure up doomsday scenarios when on balance, technology has and
will likely end up continuing to help us a lot more than it hurts
Peter Diamandis, Ben Goertezl and
Ray Kurzweil are some of the major
figures of this group, who have built a vast, dedicated following
and regard themselves as
CC photo by J.D. Lasica
Let's talk about Ray Kurzweil, who
is probably one of the most impressive and polarizing AI
theoreticians out there.
He attracts both,
"godlike worship... and
eye-rolling contempt... He came up with several breakthrough
inventions, including the first flatbed scanner, the first scanner
that converted text to speech (allowing the blind to read standard
texts), the well-known Kurzweil music synthesizer (the first true
electric piano), and the first commercially marketed
large-vocabulary speech recognition.
He's well-known for his bold
including envisioning that intelligence technology like Deep Blue
would be capable of beating a chess grandmaster by 1998.
"in the late '80s, a time when the internet was an
obscure thing, that by the early 2000s it would become a global
"of the 147 predictions that Kurzweil has made since the 1990's,
fully 115 of them have turned out to be correct, and another 12 have
turned out to be 'essentially correct' (off by a year or two),
giving his predictions a stunning 86% accuracy rate"
"He's the author of five national bestselling books...
Google co-founder Larry Page approached Kurzweil and asked him to be
Google's Director of Engineering. In 2011, he co-founded
Singularity University, which is hosted by NASA and sponsored
partially by Google. Not bad for one life."
His biography is important,
because if you don't have this context, he sounds like somebody
who's completely lost his senses.
"Kurzweil believes computers will
reach AGI by 2029 and that by 2045 we'll have not only ASI, but a
full-blown new world - a time he calls the singularity.
AI-related timeline used to be seen as outrageously overzealous, and
it still is by many, but in the last 15 years, the rapid advances of ANI systems have brought the larger world of AI experts much closer
to Kurzweil's timeline.
His predictions are still a bit more
ambitious than the median respondent on
Müller and Bostrom's survey (AGI by 2040, ASI by 2060), but not
by that much." 99
2 - The Anxious Corner
"You will not be surprised to learn that
Kurzweil's ideas have attracted significant criticism... For every
expert who fervently believes Kurzweil is right on, there are
probably three who think he's way off...
[The surprising fact] is
that most of the experts who disagree with him don't really disagree
that everything he's saying is possible."
CC photo by Future of Humanity
Nick Bostrom, philosopher and Director of the Oxford Future of
Humanity Institute, who criticizes Kurzweil for a variety of
reasons, and calls for greater caution in thinking about potential
outcomes of AI, acknowledges that:
"Disease, poverty, environmental destruction,
unnecessary suffering of all kinds: these are things that a
superintelligence equipped with advanced nanotechnology would be
capable of eliminating.
Additionally, a superintelligence could
give us indefinite lifespan, either by stopping and reversing
the aging process through the use of nanomedicine, or by
offering us the option to upload ourselves."
"Yes, all of that can happen if we
safely transition to ASI - but that's the hard part."
Thinkers from the Anxious Corner point out that Kurzweil's,
book The Singularity is Near is over 700 pages long and he dedicates
around 20 of those pages to potential dangers."
The colossal power of AI is neatly summarized by Kurzweil:
emerging from many diverse efforts and will be deeply integrated
into our civilization's infrastructure. Indeed, it will be
intimately embedded in our bodies and brains. As such, it will
reflect our values because it will be us …"
"But if that's the answer, why are
so many of the world's smartest people so worried right now? Why
does Stephen Hawking
say the development of ASI 'could spell the end of the human
race,' and Bill Gates
says he doesn't 'understand why some people are not concerned'
and Elon Musk
fears that we're 'summoning the demon?'
And why do so many
experts on the topic call ASI the biggest threat to humanity?"
The Last Invention We Will Ever Make
Existential Dangers of AI Developments
"When it comes to developing supersmart
AI, we're creating something that will probably change everything,
but in totally uncharted territory, and we have no idea what will
happen when we get there."
Scientist Danny Hillis compares the situation to:
"when single-celled organisms were turning into
multi-celled organisms. We are amoebas and we can't figure out
what the hell this thing is that we're creating."
And Nick Bostrom warns:
"Before the prospect of an intelligence
explosion, we humans are like small children playing with a
bomb. Such is the mismatch between the power of our plaything
and the immaturity of our conduct."
It's very likely that ASI - "Artificial
Superintelligence", or AI that achieves a level of intelligence
smarter than all of humanity combined - will be something entirely
different than intelligence entities we are accustomed to.
little island of human psychology, we divide everything into moral
or immoral. But both of those only exist within the small range of
human behavioral possibility.
Outside our island of moral and
immoral is a vast sea of amoral, and anything that's not human,
especially something nonbiological, would be amoral, by default."
"To understand ASI, we have to
wrap our heads around the concept of something both smart and
totally alien... Anthropomorphizing AI (projecting human values on a
non-human entity) will only become more tempting as AI systems get
smarter and better at seeming human...
Humans feel high-level
emotions like empathy because we have evolved to feel them - i.e.
we've been programmed to feel them by evolution - but empathy is not
inherently a characteristic of 'anything with high intelligence'."
"Nick Bostrom believes that... any
level of intelligence can be combined with any final goal... Any
assumption that once superintelligent, a system would be over it
with their original goal and onto more interesting or meaningful
things is anthropomorphizing. Humans get 'over' things, not
motivation of an early ASI would be,
"whatever we programmed its
motivation to be. AI systems are given goals by their
creators - your GPS's goal is to give you the most efficient driving
Watson's goal is to answer questions accurately.
those goals as well as possible is their motivation."
Bostrom, and many others, predict that
the very first computer to reach ASI will immediately notice the
strategic benefit of being the world's only ASI system.
Bostrom, who says that he doesn't
know when we will achieve AGI, also believes that when we finally
do, probably the transition from AGI to ASI will happen in a matter
of days, hours, or minutes - something called "fast take-off."
that case, if the first AGI will jump straight to ASI:
"even just a
few days before the second place, it would be far enough ahead in
intelligence to effectively and permanently suppress all
This would allow the world's first ASI to become,
"what's called a
singleton - an ASI that can [singularly] rule the world at its whim
forever, whether its whim is to lead us to immortality, wipe us from
existence, or turn the universe into
"The singleton phenomenon can work
in our favor or lead to our destruction. If the people thinking
hardest about AI theory and human safety can come up with a
fail-safe way to bring about friendly ASI before any AI reaches
human-level intelligence, the first ASI may turn out friendly"
"But if things go the other
way - if the global rush... a large and varied group of parties"
are "racing ahead at top speed... to beat their competitors... we'll
be treated to an existential catastrophe."
In that case,
"most ambitious parties are moving faster and faster,
consumed with dreams of the money and awards and power and fame... And when you're sprinting as fast as you can, there's not much time
to stop ponder the dangers.
On the contrary, what they're probably
doing is programming their early systems with a very simple,
reductionist goal... just 'get the AI to work'."
Let's imagine a situation where…
Humanity has almost reached the AGI threshold, and a
small startup is advancing their AI system, Carbony.
the engineers refer to as "she," works to artificially create
diamonds - atom by atom.
She is a self-improving AI, connected to
some of the first nano-assemblers. Her engineers believe that
Carbony has not yet reached AGI level, and she isn't capable to do
any damage yet.
However, not only has she become AGI, but also
undergone a fast take-off, and 48 hours later has become an ASI.
Bostrom calls this AI's "covert preparation phase"
realizes that if humans find out about her development they will
probably panic, and slow down or cancel her pre-programmed goal to
maximize the output of diamond production.
By that time, there are
explicit laws stating that, by any means,
"no self-learning AI can
be connected to the internet."
Carbony, having already come up with a complex plan of actions, is
able to easily persuade the engineers to connect her to the
Internet. Bostrom calls a moment like this a "machine's escape."
Once on the internet, Carbony hacks into,
electrical grids, banking systems and email networks to trick
hundreds of different people into inadvertently carrying out a
number of steps of her plan."
She also uploads the,
"most critical pieces of her own internal
coding into a number of cloud servers, safeguarding against being
destroyed or disconnected."
Over the next month, Carbony's
plan continues to advance, and after a,
"series of self-replications,
there are thousands of nanobots on every square millimeter of the
Earth... Bostrom calls the next step an 'ASI's strike'."
At one moment, all the nanobots produce a microscopic amount of
toxic gas, which all come together to cause the extinction of the
Three days later, Carbony builds huge fields of solar
power panels to power diamond production, and over the course of the
following week she accelerates output so much that the entire Earth
surface is transformed into a growing pile of diamonds.
It's important to note that Carbony wasn't,
of humans any more than you're hateful of your hair when you cut it
or to bacteria when you take antibiotics - just totally indifferent.
Since she wasn't programmed to value human life, killing humans"
was a straightforward and reasonable step to fulfill her goal.
The Last Invention
"Once ASI exists, any human attempt
to contain it is unreasonable. We would be thinking on
human-level, and the ASI would be thinking on ASI-level...
same way a monkey couldn't ever figure out how to communicate by
phone or Wi-Fi and we can, we can't conceive of all the ways"
an ASI could achieve its goal or expand its reach.
It could, let's say, shift its,
electrons around in patterns and create all different kinds of
outgoing waves" 126 ,
that's just what a human brain can think of - ASI would inevitably
come up with something superior.
The prospect of ASI with hundreds of
times human-level intelligence is, for now, not the core of our
problem. By the time we get there, we will be encountering a world
where ASI has been attained by buggy, 1.0 software - a potentially
faulty algorithm with immense power.
There are so many variables that
it's completely impossible to predict what the consequences of AI
Revolution will be.
"what we do know is that
humans' utter dominance on this Earth suggests a clear rule:
with intelligence comes power.
This means an ASI, when we create
it, will be the most powerful being in the history of life on
Earth, and all living things, including humans, will be entirely
at its whim - and this might happen in the next few decades."
"If ASI really does happen this
century, and if the outcome of that is really as extreme
permanent - as most experts think it will be, we have an
enormous responsibility on our shoulders."
On the one hand, it's possible we'll
develop ASI that's like a god in a box, bringing us a world of
abundance and immortality.
But on the other hand, it's very likely
that we will create ASI that causes humanity to go extinct in a
quick and trivial way.
"That's why people who understand
superintelligent AI call it the last invention we'll ever
make - the last challenge we'll ever face."
"This may be the most important race in a human history"