Artificial General Intelligence (AGI) is defined as AI capable of performing any intellectual task a human can, a benchmark that current AI, including advanced large language models, has not yet reached despite their impressive capabilities.
Mind Map
클릭해서 펼치기
클릭해서 인터랙티브 마인드맵 전체 보기
AGI artificial general intelligence is
an exciting concept I think some of the
confusion around it stems from the use
of the word General as you know AI is a
general purpose technology meaning that
is useful for many different things and
the rise of large language models has
led to single models like chat GPT that
are useful for many things and feel like
they could be general purpose but a
general purpose purp technology is not
the same thing as artificial general
intelligence let's take a look at what
really is the technical definition of
AGI the most widely accepted definition
of AGI is AI that could do any
intellectual task that a human can some
definitions actually say any Tas that a
human or mammal can but I'll stick with
this one for now so for example if we
have AGI AI would be able to learn to
drive a car through about 20 hours of
practice similar to what a teenager can
this is an example raised by Deep
learning Pioneer yand and today
self-driving cars still aren't quite
there yet certainly not after so little
as 20 hours of practice if we had AGI AI
would also be able to do the
intellectual task of completing a PhD
thesis level research after five years
of work or maybe even faster and today
AI can help with some parts of
brainstorming and writing and maybe be a
thought partner for some elements of
research but we're clearly very far away
from this still if we had AGI AI would
also be able to do pretty much all the
TSS of a computer programmer or really
any other knowledge workers whose
contributions to work are through carry
out intellectual toss and clearly we're
still very far away from this I know
that the different views about how long
it would take to get to AGI I think
we're still many decades of away maybe
even longer but I hope we'll get there
sometime in our lifetimes there are some
businesses that have made much more
optimistic forec cost of when we'll get
there but I found that most of those
businesses have changed the definition
of AGI and set a much lower bar to get
there I showed the definition of AGI
used by one of these businesses to an
economist friend of mine and he equipped
boy if that's the definition of AGI I
think we got there 30 years ago so it is
true that by lowering the boss sufficiently
sufficiently
we could get there much faster but for
the most widely accepted definition of
AGI I think we're still quite a long
ways away one of the exciting things
about large language models is that we
can use them as reasoning engines as I
mentioned last week and maybe we're now
starting to see what a rough outline of
what AGI could be
someday I don't think there's any
fundamental laws of physics that
prevents us from ever creating AGI which
I think will actually be very valuable
to Human Society but we'll still need
some significant technical breakthroughs
to get there one of the hard things
about getting to aggi is that it
benchmarks artificial intelligence
against human
intelligence artificial intelligence and
biological intelligence have progressed
along two very different paths for
example AI learning from far more text
than any human can read in a lifetime so
AI is already far better than any human
at certain task already but to ask AI to
do everything all intellectual tasks
that humans can do that's still just a
very high bar but even though we're
still aways from
AGI AI is very powerful and it's
important that we use it responsibly
let's take a look in the next video at responsible
responsible AI
텍스트나 타임스탬프를 클릭하면 동영상의 해당 장면으로 바로 이동합니다
공유:
대부분의 자막은 5초 이내에 준비됩니다
원클릭 복사125개 이상의 언어내용 검색타임스탬프로 이동
YouTube URL 붙여넣기
YouTube 동영상 링크를 입력하면 전체 자막을 가져옵니다
자막 추출 양식
대부분의 자막은 5초 이내에 준비됩니다
Chrome 확장 프로그램 설치
YouTube를 떠나지 않고 자막을 즉시 가져오세요. Chrome 확장 프로그램을 설치하면 동영상 시청 페이지에서 바로 자막에 원클릭으로 접근할 수 있습니다.
