[00:00]
humans rule Earth without competition
[00:03]
but we're about to create something that
[00:05]
may change that our last invention the
[00:08]
most powerful tool weapon or maybe even
[00:10]
entity artificial super
[00:13]
intelligence this sounds like science
[00:16]
fiction so let's start at the beginning
[00:19]
intelligence is the ability to learn
[00:22]
reason acquire Knowledge and Skills and
[00:24]
use them to solve problems intelligence
[00:27]
is power and we're the species that
[00:29]
exploited it the most so much so that
[00:32]
Humanity broke the game of Nature and
[00:35]
took control but the journey there
[00:37]
wasn't straightforward for most animals
[00:40]
intelligence costs too much energy to be
[00:42]
worth it still if we track intelligence
[00:45]
in the tree of species over time we can
[00:47]
see lots of diverse forms of
[00:49]
intelligence emerge the earliest brains
[00:52]
were in flatworms 500 million years ago
[00:55]
just a tiny cluster of neurons to handle
[00:57]
basic body functions it took hundreds of
[01:00]
millions of years for species to
[01:01]
diversify and become more complex life
[01:05]
conquered New environments gained new
[01:07]
senses and had to contend with Fierce
[01:09]
competition over resources but in nature
[01:12]
all that matters is survival and brains
[01:15]
are expensive so for almost all animals
[01:17]
a narrow intelligence fit for a narrow
[01:19]
range of tasks was enough in some
[01:22]
environments animals like birds
[01:24]
octopuses and mammals evolved more
[01:26]
complex neural structures for them it
[01:29]
paid off to have more energy consuming
[01:31]
skills like Advanced navigation and
[01:33]
communication until 7 billion years ago
[01:36]
the hominins emerged we don't know why
[01:39]
but their brains grew faster than their
[01:41]
relatives something was different about
[01:43]
their intelligence very slowly it turned
[01:46]
from narrow to General from a
[01:48]
screwdriver to a multi-tool able to
[01:51]
think about diverse Problems 2 million
[01:55]
years ago homo erector saw the world
[01:57]
differently from anyone before as
[01:59]
something be understood and transformed
[02:02]
they controlled fire invented tools and
[02:05]
created the first culture we probably
[02:08]
emerged from them around 250,000 years
[02:11]
ago with an even larger and more complex
[02:14]
brain it enabled us to work together in
[02:16]
large groups and to communicate complex
[02:19]
thoughts we used our intelligence to
[02:21]
improve our lives to ask how things work
[02:24]
and why things are the way they are with
[02:26]
each Discovery we asked more questions
[02:29]
and pushed forward Ward preserving what
[02:31]
we learned outpacing what evolution
[02:33]
could do with genes knowledge Builds on
[02:36]
knowledge progress was slow at first and
[02:39]
then sped up exponentially agriculture
[02:42]
writing medicine astronomy or philosophy
[02:44]
exploded into the world 200 years ago
[02:47]
science took off and made us even better
[02:49]
at learning about the world and speeding
[02:51]
up progress 35 years ago the internet
[02:54]
age began today we live in a world made
[02:57]
to suit our needs created by us us for
[03:00]
us this is incredibly new we forget how
[03:04]
hard it was to get here how enormous the
[03:07]
steps on the intelligence ladder were
[03:09]
and how long it took to climb them but
[03:11]
once we did we became the most powerful
[03:14]
animal in the world in a
[03:16]
heartbeat but we may be in the process
[03:18]
of changing this we're building machines
[03:21]
that could be better at the very thing
[03:23]
that gave us the power to conquer the
[03:25]
planet Humanity's final invention
[03:30]
artificial
[03:31]
intelligence artificial intelligence or
[03:34]
AI is software that performs mental
[03:36]
tasks with a computer code that uses
[03:38]
silicon instead of neurons to solve
[03:41]
problems in the beginning AI was very
[03:43]
simple lines of code on paper mere
[03:46]
proofs of concept to demonstrate how
[03:48]
machines could perform mental tasks only
[03:51]
in the 1960s did we start seeing the
[03:53]
first examples of what we would
[03:55]
recognize as AI a chatbot in 1964
[03:59]
approach program to sort through
[04:00]
molecules in
[04:02]
1965 slow specialized systems requiring
[04:06]
experts to use them their intelligence
[04:08]
was extremely narrow built for a single
[04:11]
task inside a controlled environment the
[04:13]
equivalent of flat worms 500 million
[04:16]
years ago doing the minimum amount of
[04:18]
mental work progress in AI research
[04:21]
paused several times when researchers
[04:23]
lost hope in the technology but just
[04:25]
like changing environments create new
[04:27]
niches for Life the world around AI
[04:30]
changed between 1950 and 2000 computers
[04:33]
got a billion times faster while
[04:35]
programming became easier and widespread
[04:38]
in 1972 AI could navigate a room in 1989
[04:42]
it could read handwritten numbers but it
[04:45]
remained a fancy tool no match for
[04:47]
humans until in 1997 an AI shocked the
[04:51]
World by beating the world champion in
[04:53]
chess proving that we could build
[04:55]
machines that could surpass us but we
[04:57]
calmed ourselves because a chest spot is
[05:00]
quite stupid not a flatworm but maybe a
[05:03]
bee only able to perform a specialized
[05:06]
narrow task but within this narrow task
[05:09]
it's so good that no human will ever
[05:11]
again beat AI a chess as computers
[05:15]
continued to improve AI became a
[05:17]
powerful tool for more and more tasks in
[05:20]
2004 it drove a robot on Mars in 2011 it
[05:23]
began recommending YouTube videos to you
[05:26]
but this was only possible because
[05:28]
humans broke down problems into easyto
[05:30]
digest chunks that computers could solve
[05:32]
quickly until we taught AIS to teach
[05:37]
themselves rise of the self-learning
[05:41]
Machines this is not a technical video
[05:43]
so we're massively oversimplifying here
[05:46]
in a nutshell the sheer power of
[05:47]
supercomputers was combined with the
[05:49]
almost endless data collected in the
[05:51]
information age to make a new generation
[05:54]
of ai ai experts began drastically
[05:57]
improving forms of AI software called
[05:59]
neur neural networks enormously huge
[06:01]
networks of artificial neurons that
[06:03]
start out being bad at their tasks they
[06:06]
then used machine learning which is an
[06:08]
umbrella term for many different
[06:10]
training techniques and environments
[06:11]
that allows algorithms to write their
[06:13]
own code and improve themselves the
[06:16]
scary thing is that we don't exactly
[06:17]
know how they do it and what happens
[06:20]
inside them just that it works and that
[06:23]
what comes out the other end is a new
[06:25]
type of AI a capable black box of code
[06:29]
the these new AIS could Master complex
[06:31]
skills extremely quickly with much less
[06:34]
human help they were still narrow
[06:36]
intelligences but a huge step up in 2014
[06:40]
Facebook AI could identify faces with
[06:42]
97% accuracy in 2016 an AI beat the best
[06:47]
humans in the incredibly complex game of
[06:49]
Go in 2018 a self-learning AI learned
[06:53]
chess in 4 hours just by playing against
[06:56]
itself and then defeated the best
[06:58]
specialized chess bot since then machine
[07:01]
learning has been applied to reading
[07:03]
image processing solving tests and much
[07:06]
more many of these AIS are already
[07:08]
better than humans for whatever narrow
[07:10]
task they were trained but they still
[07:12]
remained a simple tool AI still didn't
[07:15]
seem that big of a deal for most people
[07:17]
and then came the chatbot chat GPT the
[07:21]
work that went into it is massive it
[07:23]
trained on nearly everything written on
[07:24]
the internet to learn how to handle
[07:26]
language which it now does better than
[07:28]
most people it can summarize translate
[07:31]
and help with some math problems it's
[07:34]
incredibly more broad than any other
[07:36]
system just a few years ago not crushing
[07:39]
any single Benchmark but order of them
[07:41]
at once many large tech companies are
[07:44]
spending billions to build powerful
[07:46]
competitors AI is already transforming
[07:49]
customer service banking Healthcare
[07:51]
marketing copyrighting creative spaces
[07:54]
and more AI generated content has
[07:57]
already taken hold of social media
[07:59]
YouTube and news websites elections are
[08:02]
expected to be inundated by Propaganda
[08:04]
and
[08:05]
misinformation no one is sure how much
[08:07]
good or harm can come from adopting AI
[08:09]
everywhere change is scary there will be
[08:12]
winners and losers one of the biggest
[08:15]
questions governments and corporations
[08:17]
have now is how to manage the transition
[08:19]
to an AI boosted economy all these
[08:22]
potential gains or risks are just the
[08:24]
result of today's AI chat gpt's
[08:27]
intelligence is a major step up but it
[08:30]
remains narrow one it can write a great
[08:32]
essay in seconds it doesn't understand
[08:35]
what it's writing but what if the AIS
[08:38]
stopped being
[08:39]
narrow General AI what makes humans
[08:43]
different from current AI is our general
[08:46]
intelligence humans can technically
[08:48]
absorb any piece of knowledge and start
[08:50]
working on any problem we're great at
[08:52]
many very different skills and tasks
[08:54]
from playing chess to writing or solving
[08:56]
science puzzles not equally of course
[08:59]
some of us are experts in some fields
[09:01]
and beginners in others but we can
[09:03]
technically do all of them in the past
[09:06]
AI was narrow and able to become good at
[09:08]
one skill but was rather bad in all the
[09:10]
others simply by building faster
[09:13]
computers and pouring more money into AI
[09:15]
training will get us new more powerful
[09:18]
generations of AI but what is the next
[09:21]
step for AI is to become a general
[09:24]
intelligence like us an
[09:26]
AGI if the AI improvement process
[09:29]
continues as it has been it's not
[09:31]
unlikely that AGI could be better in
[09:33]
most or even all skills that humans can
[09:36]
do we don't know how to build AGI how it
[09:39]
will work or what it will be able to do
[09:42]
since narrow AIS today are capable of
[09:45]
mastering one mental task quickly HGI
[09:47]
might be able to do the same with all
[09:49]
mental tasks so even if it starts out
[09:52]
stupid and HGI might be able to become
[09:55]
as smart and capable as a human while
[09:58]
this sounds like science sence fiction
[10:00]
most AI researchers think this will
[10:02]
happen sometime this Century maybe
[10:04]
already in a few years humanity is not
[10:07]
ready for what will happen next not
[10:09]
socially not economically not morally
[10:13]
earlier we defined intelligence as the
[10:15]
ability to learn reason acquire
[10:17]
Knowledge and Skills and use them to
[10:19]
solve problems all things humans excel
[10:22]
at an AGI as intelligent as even an
[10:25]
average human would already disrupt
[10:27]
modern civilization because they're not
[10:29]
Bound by the same limitations as we are
[10:32]
today's AIS like chat gbt already think
[10:35]
and solve the tasks they were made for
[10:37]
at least 10 times faster than even very
[10:39]
skilled humans maybe AGI will be slower
[10:43]
but it may also be faster maybe much
[10:45]
faster and since HGI are software you
[10:49]
could copy them endlessly as long as you
[10:51]
have enough storage and run them in
[10:54]
parallel there are 8 million scientists
[10:56]
in the world now imagine an AI copied a
[10:59]
million times and put to work imagine 1
[11:02]
million scientists working 24/7 thinking
[11:05]
10 times faster than humans without
[11:07]
being distracted only focused on the
[11:10]
task they've been given what if suddenly
[11:13]
HGI could do all intelligence-based jobs
[11:16]
in the world from interpreting law to
[11:18]
coding to creating animated YouTube
[11:20]
videos better faster and much cheaper
[11:23]
than humans would whoever controls this
[11:26]
AGI suddenly own the economy
[11:29]
and thinking bigger human progress is
[11:32]
our intelligence applied to problems so
[11:35]
what could a million agis
[11:37]
achieve solve fundamental questions of
[11:40]
science like dark energy invent new
[11:43]
technology that gives us Limitless
[11:44]
energy fix climate change cure aging and
[11:48]
cancer but then again sadly humans apply
[11:52]
their intelligence not just for the
[11:53]
benefit of all what if the agis are
[11:56]
tasked to guide drones or pull the
[11:58]
triggers in war or to engineer a virus
[12:01]
that only kills people with green eyes
[12:04]
or to create the most profitable social
[12:06]
media so addictive that people starve in
[12:08]
front of their screens the creation of
[12:11]
AGI could reasonably be as big of an
[12:13]
event as taming fire or electricity and
[12:16]
give whoever invents it equally as much
[12:19]
power but now let's go one step further
[12:23]
what if the potential of AGI doesn't
[12:25]
stop here intelligence explosion
[12:29]
intelligence and knowledge build and
[12:31]
accelerate each other but humans are
[12:33]
limited by biology and evolution once we
[12:37]
evolved the right Hardware our software
[12:39]
outpaced evolution by orders of
[12:41]
magnitudes and within a heartbeat we
[12:43]
ruled this planet but our software
[12:46]
basically hasn't changed much since then
[12:48]
which is why we have obesity and destroy
[12:50]
the climate for short-term gains since
[12:53]
AGI is software on a computer once it's
[12:56]
smart enough to do AI research the rate
[12:58]
of AI progress should speed up a lot and
[13:01]
that results in better AI That's better
[13:04]
at AI research without much human
[13:06]
involvement it may even be possible that
[13:08]
AI could learn how to directly improve
[13:10]
itself in which case some experts fear
[13:12]
this feedback loop could be incredibly
[13:15]
fast maybe just months or years after
[13:18]
the first self-improving hii is switched
[13:20]
on maybe it would actually take decades
[13:23]
we simply don't know this is all
[13:25]
speculative but such an intelligence
[13:27]
explosion might lead to a true
[13:29]
superintelligent entity we don't know
[13:32]
what such a being would look like what
[13:33]
its motives or goals would be what would
[13:36]
go on in its inner world we could be as
[13:39]
laughably stupid to superintelligence as
[13:42]
squirrels are to us unable to even
[13:45]
comprehend its way of
[13:46]
thinking this hypothetical scenario
[13:49]
keeps many people up at night humanity
[13:52]
is the only example we have of an animal
[13:55]
becoming smarter than all others and we
[13:57]
have not been kind to what we perceive
[14:00]
as less intelligent beings AGI might be
[14:03]
the last invention of humanity it's
[14:06]
possible that it could become the most
[14:08]
intelligent and therefore most powerful
[14:10]
being on earth a God in a box that could
[14:13]
exercise its power to bring unimaginable
[14:16]
wealth and happiness to humans while
[14:18]
securing our future or it could subvert
[14:21]
civilization and bring about our end
[14:24]
with Humanity unable to come up with a
[14:26]
way to stop it we'll look at some of
[14:28]
these potential Futures in more videos
[14:30]
but for now let's wrap up the only thing
[14:33]
we know for sure is that today right now
[14:36]
many of the largest and richest
[14:38]
companies in the world are racing to
[14:40]
create ever more powerful AIS whatever
[14:43]
our future is we are running towards
[14:48]
it who knows how long we have until we
[14:51]
must confront our AI future luckily you
[14:54]
still have plenty of time to prepare for
[14:56]
it if you're learning on Brilliance that
[14:58]
is
[14:59]
brilliant will make you a better thinker
[15:01]
and Problem Solver in just minutes a day
[15:04]
with thousands of bite-sized Hands-On
[15:06]
lessons on just about anything you may
[15:08]
be curious about including AI their
[15:11]
latest course how llms work takes you
[15:14]
under the hood of real language models
[15:17]
it demystifies Technologies like chat
[15:19]
gbt with interactive lessons on
[15:21]
everything from how models build
[15:23]
vocabulary to how they choose their next
[15:25]
word you'll learn how to tune llms to
[15:27]
produce output with exactly your desired
[15:29]
tonality whether it's poetry or a cover
[15:32]
letter and you'll understand why
[15:34]
training is really everything by
[15:36]
comparing models trained on Taylor Swift
[15:38]
lyrics and the legal speech of big Tech
[15:40]
terms and conditions it's an immersive
[15:42]
AI Workshop allowing you to experience
[15:45]
and harness the mechanics of today's
[15:46]
most advanced tool we've also partnered
[15:49]
with brilliant to create a series of
[15:51]
lessons to take your scientific
[15:52]
knowledge to the next level these
[15:54]
lessons let you further explore the
[15:56]
topics in our most popular videos from
[15:58]
rapes and mum metabolism to climate
[16:00]
science and supern noi each lesson on
[16:03]
brilliant is interactive like a
[16:04]
one-on-one version of a cazar video and
[16:07]
you can get started Whenever Wherever
[16:10]
right from whatever device you'd like to
[16:12]
get handson with ktz kazak lessons and
[16:14]
explore everything brilliant has to
[16:15]
offer from Ai and programming to maths
[16:18]
science and Beyond start your free
[16:20]
30-day trial by signing up at
[16:22]
brilliant.org
[16:23]
nutshell there's even an extra perk for
[16:25]
courts gazar viewers anyone signing up
[16:27]
through our link will get get 20% off an
[16:30]
annual membership once their trial ends
[16:33]
[Music]