00:00:00
my videos about new trends always go
00:00:03
badly I think it's because if something
00:00:06
isn't much talked about already it
00:00:08
doesn't seem all that interesting but it
00:00:11
makes me so proud if I can later say see
00:00:14
I've been talking about this for years
00:00:16
today's video is one of those I want to
00:00:19
explain why a currently relatively
00:00:21
unknown approach to Quantum Computing
00:00:23
that hasn't gotten very far yet might
00:00:26
pull ahead very quickly thanks to
00:00:29
artificial intelligence artificial
00:00:31
intelligence and Quantum Computing are
00:00:33
the two currently most exciting
00:00:35
technological developments we've
00:00:38
previously discussed how they're
00:00:39
competing and how AI is winning so far
00:00:43
but in the past year we've seen another
00:00:45
Trend that's using AI to make quantum
00:00:49
computers work this has recently been
00:00:51
applied to an approach called atoms in
00:00:54
tweezers with remarkable success Quantum
00:00:57
Computing has been heralded and the next
00:01:00
Industrial Revolution and the future of
00:01:03
computing a technology that can
00:01:05
allegedly do anything from curing cancer
00:01:08
to fixing climate change to resurrecting
00:01:10
Elvis Presley I made up the last one but
00:01:13
the way things are going I wouldn't be
00:01:15
surprised if that would be next week's
00:01:17
headline what is true though is that
00:01:19
quantum computers could vastly speed up
00:01:22
the calculations necessary for financial
00:01:25
optimizations that means you could use
00:01:28
them to make money on the stock market
00:01:29
Market loads of money and that I think
00:01:33
explains a lot of the interest in
00:01:35
Quantum Computing it's no coincidence
00:01:38
that Wells Fargo City group and HSBC are
00:01:42
investing in Quantum Computing at himim
00:01:44
Israel from Bank of America
00:01:47
enthusiastically proclaimed that Quantum
00:01:49
Computing will be bigger than fire it's
00:01:52
not because he wants to cure cancer or
00:01:54
fixed climate change it's because Shing
00:01:56
us cat is more interesting when it's
00:01:59
made of gold the trouble with Quantum
00:02:01
Computing is that the current devices
00:02:03
are far away from the sizes necessary to
00:02:06
do any commercially relevant
00:02:08
calculations and the problem isn't just
00:02:10
to make them bigger the problem is to
00:02:12
make them bigger while keeping the
00:02:15
number of Errors down at least this used
00:02:18
to be the case but artificial
00:02:19
intelligence is now getting so good it's
00:02:22
learning to control the errors one
00:02:25
impressive recent example for how to use
00:02:27
AI for error correction comes from the
00:02:30
Google AI group they've trained their AI
00:02:33
on data from the Sycamore chip so that
00:02:35
superc conducting cubits then they use
00:02:38
the AI for active feedback to reduce the
00:02:41
error rate you see the results of the
00:02:43
new algorithm here in the red curves
00:02:45
it's significantly better they call
00:02:47
their system Alpha cubid and demonstrate
00:02:50
that it outperforms all other known
00:02:52
algorithms when I say it's significantly
00:02:54
better I mean it's a measurable
00:02:56
difference not a large one but it
00:02:59
generalizes better beyond the training
00:03:01
set to longer computations than previous
00:03:03
methods which will make it useful for
00:03:06
scaling up quantum computers personally
00:03:08
I don't think the superconducting cubits
00:03:11
that Google uses will bring us to
00:03:13
commercially relevant quantum computers
00:03:15
but another recent development might get
00:03:18
us there it comes from building quantum
00:03:20
computers using an approach called atoms
00:03:23
in tweezers or neutral atom arrays we've
00:03:27
talked about this previously because
00:03:29
it's one of the fastest developing
00:03:31
newcomers in Quantum Computing for this
00:03:34
one uses electromagnetic fields to trap
00:03:37
atoms in an array and entangle them with
00:03:40
lasers in traps that look like this the
00:03:43
good thing about this is that because
00:03:45
atoms are small you can pack many of
00:03:48
them closely together so this approach
00:03:51
could scale very quickly a recent survey
00:03:53
among Quantum Computing experts found
00:03:56
they rated neutral atoms as the approach
00:03:59
which holds the most promise for
00:04:02
achieving practical applications of
00:04:04
quantum Computing then again this survey
00:04:06
was done by company which works on
00:04:09
neutral atoms so take this with a pound
00:04:12
of salt one of the issues with neutral
00:04:14
atoms is though that you need them to
00:04:17
fill in the aray in an orderly way in
00:04:20
this new paper now they used AI to learn
00:04:22
how to read out the positioning of the
00:04:25
atoms and to move them into place which
00:04:28
worked like a charm in this amazing atom
00:04:30
tomography image each dot is an atom
00:04:33
there are more than 2,000 atoms of them
00:04:36
in an almost perfect array each could
00:04:39
act as a cubit for comparison the
00:04:41
currently largest quantum computer is
00:04:43
from IBM and it has about 1,000 cubits
00:04:47
let me be clear though that they haven't
00:04:49
actually done a calculation with these
00:04:51
atoms and tweezers yet but this method
00:04:54
has been catching up quickly and I
00:04:57
expect we'll hear more about this soon
00:04:59
this isn't just interesting for
00:05:01
practical purposes because it'll make
00:05:04
scating quantum computers easier I also
00:05:06
think this is interesting from a
00:05:08
research perspective because it's
00:05:11
implicitly testing how random quantum
00:05:14
mechanics really is physicists currently
00:05:16
think that quantum mechanics has this
00:05:18
irreducible random element but what if
00:05:21
it hasn't then AI might become better at
00:05:24
eliminating errors than should be
00:05:26
theoretically possible now wouldn't that
00:05:29
be interesting artificial intelligence
00:05:31
is really everywhere these days if you
00:05:34
want to learn more about how neuron
00:05:36
networks and large language models work
00:05:39
I recommend you check out the courses on
00:05:41
brilliant all courses on brilliant have
00:05:44
interactive visualizations and come with
00:05:46
follow-up questions what you see here is
00:05:48
from their data science courses which
00:05:51
they just released they all use real
00:05:53
world examples like what it means to go
00:05:56
viral on X brilliant covers a large
00:05:58
variety of topics I and science computer
00:06:00
science and maths from General
00:06:02
scientific thinking to Dedicated courses
00:06:05
on algebra or large language models just
00:06:08
what I'm interested in and they're
00:06:10
adding new courses each month I even
00:06:13
have my own course on brilliant that's
00:06:15
an introduction to Quantum Mechanics
00:06:17
it'll help you understand what a wave
00:06:19
function is and what the differences
00:06:21
between superpositions and entanglement
00:06:24
it also covers interference the
00:06:26
uncertainty principle and B theorem and
00:06:29
after that you can continue maybe with a
00:06:32
course on Quantum Computing or
00:06:34
differential equations and of course I
00:06:36
have a special offer for viewers of this
00:06:39
channel if you use my link brilliant.org
00:06:42
Saina or scan the QR code you'll get to
00:06:45
try out everything brilliant has to
00:06:47
offer for a full 30 days and you'll get
00:06:50
20% off the annual premium subscription
00:06:53
so go and check this out thanks for
00:06:55
watching see you tomorrow