- il y a 5 semaines
Quantum 101, Part 2 Harnessing Quantum Innovation for Business (and Social) Revolutions
Catégorie
🤖
TechnologieTranscription
00:00Oh, and we're back for Quantum 101 Part 2.
00:05In the previous panel...
00:10Can you hear me? Is that working?
00:12We are quantum...
00:14On this panel...
00:20On this panel, we're going to dive into the impact.
00:23Where is quantum going to change your life?
00:31We have some incredible guests.
00:33They're better than the microphones.
00:34And I will let them introduce themselves.
00:38Could you introduce yourselves
00:40and tell the audience what quantum computing means to you
00:44personally, professionally, culturally, financially?
00:48Any way you like, but introduce yourself
00:51and talk about your passion for quantum.
00:54And we'll start with you, Araceli.
00:56Sure. Hi, everyone. Can you hear me okay?
00:59Hi, I'm Araceli Venegas-Gómez.
01:01I'm the founder and CEO of Cureka.
01:03And since 2019, we focus on workforce skills,
01:07training, and everything that you need.
01:09But we are not really building a quantum computer.
01:11What is quantum computing for me?
01:13Well, I changed my career because I love quantum.
01:15So, yeah, it's really a passion.
01:18What I think is important for you is to know
01:20that quantum computing and quantum technologies
01:22are going to be changing our society,
01:25are going to be changing our businesses,
01:27are going to be changing the way we do a lot of things.
01:29It's going to be so disruptive
01:30that I don't think that we have the answer right now
01:33how much it's going to change everything.
01:35It's going to be amazing.
01:37But in order to prepare for that,
01:39yes, we need some time.
01:40But we need to start hearing about it.
01:42We need to start reading about it.
01:43We need to start getting ready for that.
01:45So quantum computing for me
01:47is going to be the next change in our society
01:49and in our history.
01:51I like your use of the word destructive there.
01:56Joe.
01:58Joe, number one.
02:00Sure.
02:02So I'm the founder and CEO
02:05of a company called Horizon Quantum Computing,
02:07but I've been working in the area
02:09for about 20 years at this point.
02:12I got into quantum computing
02:15because when I was growing up,
02:17I thought it would be really cool
02:18to be part of a technology revolution,
02:21to have been there in the first days of computers
02:25or the first days of PCs
02:27or the first days of the internet.
02:28So I made the bet about 20 years ago
02:31that quantum computing was my best way of getting there.
02:35And, you know, the first couple of years
02:37was really exciting,
02:38and then it looks like the systems
02:40are a bit further off
02:41than we thought they were going to be.
02:43And, you know, it took a little while,
02:45but we're getting to a point now
02:46where quantum computing
02:48is starting to become a reality,
02:49where we're starting to see progress
02:52towards error-corrected quantum computers,
02:54towards quantum computers
02:55that can outperform conventional computers.
02:58We're not quite there yet,
03:00but we're starting to get to the point
03:02where that's possible.
03:04We're just on the verge of starting to see this.
03:08So for me, really, what quantum computing is,
03:11it's the potential to be computing all over again.
03:14Now, that's a big guess,
03:17and maybe it's not,
03:19but that's the promise of it for me.
03:21Excellent, thank you.
03:22And Joe, number one, as well,
03:25because this is the supposition of Joe.
03:28We do.
03:30There are two of us on stage.
03:31Hi, I'm Joe Brose.
03:33I'm from IBM,
03:35and I'm vice president of IBM Quantum,
03:38specifically in the growth and marketing area
03:41of IBM Quantum.
03:43And what drew me to quantum initially
03:47is the physics of quantum computing is beautiful.
03:53I'm a physicist by training,
03:56and the physics involved in a quantum computer
04:00is just very, very compelling.
04:04But as time has gone on,
04:07that's shifted a little bit
04:08because now we have this tool.
04:11and what is really interesting to me now
04:15is how to utilize that tool
04:18as quantum is maturing.
04:21How do we utilize that tool?
04:23So initially, it was kind of around
04:27the physics for quantum,
04:30and now that's shifted,
04:31and that is quantum for physics
04:34and other applications.
04:36How do we use it?
04:37How do we do things with quantum computers?
04:42Thank you for your honesty.
04:45Joe Brose, I'd like to stay with you,
04:47and could you,
04:49speaking through the lens of IBM,
04:51talk about a few projects
04:53or domains that you are currently thinking about
04:56and actively working on with IBM?
04:59Yes.
05:00So, at IBM, here in Europe,
05:03we have over 90 partners,
05:06corporate partners,
05:08research and development entities,
05:10just here in Europe alone,
05:12over 250 globally.
05:15We work on many different use cases.
05:17The use cases range in everything
05:20from healthcare and life sciences
05:22to fundamental materials.
05:27We look at chemistry.
05:30We're looking at applications
05:32in the financial sector as well.
05:35Some of those applications, for example,
05:37are in protein folding.
05:40How do proteins fold?
05:42And that turns out to be
05:45a very important thing to understand
05:48when it comes to healthcare,
05:51drug discovery,
05:52how small molecules might work
05:55as medicines,
05:57as drugs,
05:58or pharmacological agents.
05:59In the finance sector,
06:03we've done quite a lot of work
06:05with companies and banks
06:07in the area of, for instance,
06:09fraud detection.
06:10How do you detect
06:12when a transaction is fraudulent?
06:16What are the features
06:17that you select for?
06:19And how do you get a high hit rate
06:23with a low false positive rate?
06:26And do that better and better
06:28so that you can detect fraud
06:30without a bunch of
06:33false positive use cases.
06:36Another area is in materials discovery.
06:39We're doing quite a lot right now
06:42in the area of spin chains,
06:47spin Hamiltonians,
06:49for various model systems
06:53in condensed matter physics.
06:55So those are widely disparate use cases,
06:59but they are all use cases
07:01that we are working on
07:02with clients at IBM Quantum.
07:06Thank you.
07:08Joe,
07:11software and hardware.
07:13So I'm going to show my ignorance here.
07:15So if quantum computing
07:17is the hardware,
07:19you're more involved
07:20in the software side of this.
07:24How does software fit
07:25into the quantum paradigm?
07:27What is Horizon doing
07:29to push it forward,
07:32to evolve it?
07:34Sure.
07:35I mean,
07:35I should say that I consider
07:36it both quantum computing,
07:38both the hardware
07:39and the software.
07:40I'm a theoretical physicist
07:42by background,
07:43so most people in the space
07:45are coming from physics backgrounds
07:47either way.
07:48In terms of how to make use
07:52of quantum computing
07:53to actually solve
07:54real-world problems with it,
07:55it's not just a question
07:56of building good quantum computers.
07:59We also need the algorithms
08:00to be able to tackle those problems.
08:02And it turns out
08:03that coming up with quantum algorithms
08:04is extremely difficult.
08:05So we build tools
08:06to try to make that easier.
08:08We want to make it easier
08:09for domain experts
08:10to be able to apply quantum computing
08:12in different fields,
08:14be that in finance,
08:15be that pharma,
08:16or be it in aerospace
08:18or automotive
08:19or the energy sector.
08:20We need to be able
08:21to empower domain experts
08:22to take advantage
08:23of the hardware.
08:24However,
08:24let me just add
08:25that tied integration
08:27between hardware and software
08:28is really critical
08:29to getting to useful
08:30quantum computing.
08:31So we've recently
08:32made the decision
08:33to start investing
08:33in quantum hardware,
08:35start operating
08:35our own quantum hardware
08:36on-premises.
08:37We're not building
08:38the QPUs ourselves,
08:39we buy those from partners,
08:40but I think we are now
08:42the first software company
08:43to have our own
08:44quantum computer
08:45so we're,
08:46I mean,
08:47looking to a future
08:48where I think
08:48that's going to become
08:49increasingly common.
08:51I wanted to ask this question
08:53and I just realized
08:53it's probably not
08:54a very good question,
08:55but should we think
08:56about software
08:57and hardware
08:58in the same way
08:59that we are familiar
09:00with software
09:01on our laptops
09:01and our computers
09:02at home?
09:03Is that a comparable
09:06way to think about it?
09:09Basically, yes.
09:11What we do at Horizon
09:12is we build
09:12software development tools
09:13so we build
09:14the compilers
09:15and the tools
09:16that you use
09:17to make
09:17and deploy
09:18software applications
09:19rather than
09:20the applications
09:21themselves,
09:21but there are
09:22definitely people
09:23who are developing
09:23applications
09:24for chemistry,
09:25for finance,
09:26for a whole range
09:27of sectors
09:27that are intended
09:28to run on quantum computers.
09:30So just as your laptop
09:33would be useless
09:34if you wiped
09:35the hard drive,
09:36your quantum computer
09:36is also not very good
09:37without software
09:38to run on it.
09:41Okay, thank you.
09:43So, there's a lot
09:46of great minds
09:47working on this
09:48but in order for it
09:49to become widespread
09:51and common
09:51and to impact our lives,
09:53we need people
09:54who aren't gigabrains
09:56to be able to use
09:57and manipulate
09:58and program
09:59and do this.
10:01Araceli, I believe
10:02that's where you come in
10:05to the picture
10:06a little bit.
10:06could you tell us
10:06about what you're doing
10:07and how you're making
10:08that more open
10:10to the world?
10:12Sure.
10:13So, if you're going
10:14to be using
10:14a new technology
10:15and someone comes
10:16to you and say,
10:17do you want
10:18to learn quantum?
10:19You have normally
10:20two reactions.
10:21One like,
10:22oh no,
10:22that's too difficult
10:23and the other one,
10:24oh wow, yes,
10:25and they start talking
10:26about black holes
10:27or things that have
10:28nothing to do
10:28with quantum.
10:29So, there is a lot
10:30of awareness.
10:30But black holes are cool.
10:31They are cool.
10:32There is a lot
10:33of awareness to do.
10:35So, first is
10:36understanding the basics.
10:38What it's all about?
10:39Why should I care
10:40about this?
10:41And then if you're
10:41a company,
10:42the second step is
10:43what is it for me?
10:45What is the business
10:46for me?
10:47What this technology
10:48can do for my business?
10:50That's the most
10:51important thing.
10:52But if you don't
10:52go through that,
10:53all what they are
10:54doing that is fantastic,
10:55you cannot apply
10:56because you don't
10:57have the awareness.
10:57You don't understand
10:58that it could be
10:59really a business
11:00game changer for you.
11:02Then the third step
11:03came in when we said,
11:04okay, now that you
11:05understand that this
11:06can change your business,
11:07let's go into a
11:08specific use case.
11:09Let's look at the
11:10problems that you
11:10have already and see
11:12how we can make
11:12your business more
11:13profitable because at
11:14the end it's all
11:15what we care about.
11:16How can we change
11:17society in a way
11:18that we make things
11:19better?
11:19And then we start
11:20using the hardware,
11:21the software,
11:21and all the other
11:22technologies.
11:23And what we do
11:24is trying to go
11:25with you,
11:26individuals,
11:26companies,
11:27and universities,
11:28through all these
11:28different stages
11:29so that you are ready
11:30once the technology
11:31is also ready.
11:33Okay.
11:34Joe Brose,
11:35what is IBM doing
11:37to bridge the gap
11:38between this,
11:40the technology
11:42of quantum computers,
11:43which is quite mind-bending
11:45for most people,
11:45and real-world application?
11:47Sure, thanks.
11:49You know,
11:49picking up on what
11:50Araceli said,
11:51our goal is to have
11:53quantum software
11:55become very transparent.
11:58Today,
12:00people earn a living
12:01in the digital society
12:02out here in this
12:04vast hall
12:04is being driven
12:06by electronics.
12:09But we don't have
12:10to think about
12:11what is a transistor,
12:14what is a band gap,
12:15how does an N-type
12:17or a P-type
12:18semiconductor work.
12:20People don't need
12:21to think about that.
12:23Good.
12:24They can just simply
12:25go online
12:26and develop
12:29for a use case.
12:30That's really
12:31what we would like
12:32to do for quantum.
12:34Forget about the qubits.
12:36How does this bring value?
12:38Can a developer
12:39transparently
12:41take advantage
12:43of quantum computing
12:44and begin to develop
12:46very efficiently
12:46and very quickly?
12:48So one of the things
12:49that we're driving for
12:51at IBM Quantum
12:52is to make sure
12:53that our software
12:56development kit
12:57and what we put on
12:59in open libraries,
13:01in pre-written functionals,
13:04that that material
13:05is very user-friendly,
13:07it's very digestible
13:08by the developer,
13:11that they can grab a hold
13:12of it
13:13and use it
13:14and also benefit
13:16from our open-source software
13:18from what other developers
13:20have developed
13:21previously,
13:22grab a hold of that
13:24in the library,
13:25use it for their application.
13:27And they don't have
13:28to think about
13:29the details
13:31of what a qubit
13:32might be,
13:33they don't have to think
13:34about the details
13:35of quantum computing,
13:36but rather
13:37use efficient libraries
13:40that are available
13:41to them
13:42and that's what
13:43we're striving for
13:44at IBM.
13:45You're speaking
13:46in words
13:46which I recognize,
13:47I'm not a developer
13:48or a coder,
13:48but I understand
13:50some of these words.
13:51So,
13:52and perhaps this is
13:52for you as well,
13:53Joe,
13:53if people here
13:54are traditional developers
13:56working in
13:58existing systems,
14:01is it easy
14:02for them
14:03to move over
14:03into quantum
14:04and how would
14:05they go about it?
14:06What would they
14:07need to learn?
14:08So,
14:09I think this is
14:10a great question
14:11and actually
14:12it's at the core
14:13of what we do
14:14at Horizon.
14:15So,
14:16Joe described
14:17one approach
14:18to enabling developers
14:19to make use
14:20of quantum computing
14:20without necessarily
14:22becoming experts
14:23in quantum computing
14:23themselves
14:24and that is
14:25to build
14:27pre-built
14:27building blocks
14:28that can be used
14:29to construct programs.
14:31So,
14:31you see this used
14:32a lot in machine learning.
14:33People don't necessarily
14:34develop the underlying
14:36linear algebra
14:36techniques themselves.
14:38They just use
14:40NumPy
14:41and they use
14:41other tools
14:42that are already
14:43pre-built.
14:43However,
14:44it's also true
14:45that some people
14:46need to be able
14:46to write these
14:47core functions.
14:48They need to be able
14:49to develop
14:50not just
14:51what the pre-written
14:53building blocks
14:53can do,
14:54but to go beyond
14:55that and to explore
14:56new applications.
14:57And that's where
14:58we come in.
14:59So,
14:59our focus has been
15:00on trying to take
15:01code written
15:02for conventional
15:03computers
15:04and automatically
15:05construct quantum
15:06algorithms.
15:07Algorithms that
15:08will run
15:08more efficiently
15:09on quantum
15:10computers
15:10from that
15:11classical code.
15:12And I think,
15:13you know,
15:14both are approaches
15:16you can take
15:16and we're yet
15:18to see really
15:19how things will
15:20play out
15:20in the long term.
15:21I think probably
15:22both parts will
15:23play an important role.
15:24And I think
15:26they're very
15:26mutually reinforcing
15:28of each other.
15:29We really want
15:30to build a platform
15:31on which
15:33the quantum economy
15:34and quantum businesses
15:36can grow
15:37on that platform.
15:39In the analogy
15:40to the iPhone
15:42or the Android,
15:43we want to build
15:45the hardware
15:45in a basic
15:46software development
15:47open source kit,
15:49but we want others
15:50to put on
15:51the applications
15:52on that platform.
15:53so let others
15:54develop the specific
15:56apps that can
15:57go on that platform.
16:02Just so you know,
16:03at the end,
16:03I'm going to open
16:04the floor for questions
16:05the last few minutes
16:05so start thinking
16:06about the questions
16:07that you have
16:08for our panelists.
16:11Araceli,
16:13when we were
16:13prepping for this,
16:15you made a very
16:16important point
16:17and quantum computing
16:19isn't just about
16:21quantum computing.
16:23The conversation,
16:25the quantum conversation
16:26is infinitely more
16:29vast than that
16:31and you got very
16:31passionate about it
16:32and the name of
16:33the talk is
16:34business and social
16:36revolutions
16:37caused by quantum.
16:38So could you just
16:39talk about
16:40where else
16:42outside of
16:44quantum computing
16:45we should be thinking
16:46and building?
16:47Sure,
16:48I can give you
16:49some examples
16:50of all the
16:50quantum technologies
16:51and some specific
16:52applications
16:53because I think
16:53that sometimes
16:54it's really
16:54how can I use it
16:55the most important
16:56thing for all of you.
16:58You probably have
16:58heard that
16:59if we have a
17:00quantum computing
17:00all our cryptography
17:02can break,
17:03all our encryption,
17:04all what we think
17:05that now is secure
17:06is at risk
17:07and that is really
17:08scary.
17:09The thing is that
17:10we know what is
17:10the problem
17:11but we also have
17:12a solution for that
17:13that surprisingly
17:14is quantum technologies
17:15as well.
17:15we have quantum
17:16communications technology
17:18the hardware and the
17:18software are not going
17:19to go into the details
17:20but we also know
17:21that using that
17:22we protect our
17:24encryption
17:24so don't worry
17:25everything is going
17:25to be fine.
17:26We also have
17:27quantum sensors
17:28that can be applied
17:29in a variety
17:30of use cases
17:31and different
17:33business sectors
17:33that I don't think
17:34people know about.
17:35I'm going to give you
17:36two use cases
17:37that I think
17:38are really interesting.
17:40One is the
17:40construction sector
17:41I think everyone knows
17:42that we need to
17:43build more buildings
17:44right?
17:45every day
17:46people are going
17:47in a very big city
17:49they open a hole
17:50and then they realize
17:51that oops
17:51we cannot actually
17:53build here
17:53they need to close
17:54the hole
17:54and go and move
17:55somewhere else.
17:56You could imagine
17:58how many millions
17:59every construction
18:00company
18:01lose every year
18:02because they need
18:03to close the hole
18:04and move somewhere else.
18:05So there are actually
18:05quantum sensors
18:06that can make
18:07let's say
18:08a radiography
18:09of the magnetic field
18:11and you can know
18:12what is on the ground
18:13before you make the hole.
18:14again
18:15this is really
18:16high level
18:16but you could
18:17understand
18:17that this is
18:17a very good
18:19business case
18:19for the construction
18:20companies.
18:21Another one is
18:22you all know
18:23that the dairy
18:25sector
18:26so milk producers
18:27are making millions
18:28every year.
18:29It's a huge
18:30business sector.
18:31If you know
18:32when the cow
18:33is more happy
18:34they are going
18:34to produce more milk
18:35then you are going
18:36to make more millions
18:37out of it.
18:38So there are
18:39scientific developments
18:41now where you can
18:42measure the magnetic
18:43field of the heart
18:44of the cows
18:45to know when the cows
18:46are more happy
18:47so you know
18:47how to make the cow
18:48happier
18:48then you have
18:49more milk
18:49then you have
18:50more business.
18:51And these are just
18:51two specific use cases
18:53that I think a lot
18:54of people don't even
18:54have in mind.
18:56So imagine
18:56in all the business
18:58sectors
18:58how disruptive
18:59this technology
19:00could be.
19:02I'm speechless.
19:06Happy cows.
19:08Anything you'd like
19:09to add on
19:10quantum sensors
19:11and these other
19:12applications
19:12of the technology
19:14Joe, Rose?
19:16Yes.
19:17So IBM
19:18is a quantum
19:19computing company
19:20but you know
19:22we're quite aware
19:23of what's going
19:24on in the quantum
19:25sensing world.
19:26I've worked
19:26personally in that
19:28area.
19:29So quantum sensors
19:31offer tremendous
19:32promise from the
19:34standpoint of their
19:35sensitivity.
19:36the other thing
19:38that I mean
19:39they are much
19:40closer in
19:42they're already
19:43commercial
19:43you can buy
19:44quantum sensors
19:46those are available
19:48and they see
19:51below a standard
19:52quantum limit.
19:53So normally
19:54sensors are
19:56limited to
19:57the standard
19:58quantum limit
19:58derived by
20:00what's known
20:00as shot noise.
20:01Quantum sensors
20:03do not suffer
20:04from that limitation
20:05and have much
20:06lower sensitivity
20:08levels to what
20:09is known as
20:10the Heisenberg
20:11limit.
20:11It's a quantum
20:12effect that
20:14allows you to
20:15see at a much
20:16better sensitivity
20:18using quantum
20:19sensors.
20:20There are sensors
20:21for gravity
20:23there are sensors
20:24for magnetic
20:25fields.
20:25You can imagine
20:26the happy cows
20:28and other use
20:30cases for
20:31magnetic fields.
20:33encephalography
20:34and
20:36cardiology.
20:37But also
20:39quantum sensors
20:40are used for
20:41accelerometers
20:41and
20:43gyroscopes
20:44and a whole
20:45host of use
20:46cases that are
20:48societally
20:49impactful
20:50and I think
20:51very important.
20:53Although
20:53again
20:54at IBM
20:55quantum
20:55we don't deal
20:56with quantum
20:57sensors
20:57we are certainly
20:58aware of
20:59their uses.
21:01I wonder if
21:02there's anything
21:02quantum can't
21:03do.
21:04Joe?
21:04So who
21:06you know
21:06I agree with
21:08what the others
21:09have said
21:09about the
21:09potential of
21:10quantum sensors
21:11and quantum
21:12communications
21:13for quantum
21:14cryptography.
21:16what I would
21:16say however
21:17is
21:18all of
21:19these
21:19essentially
21:20go under
21:21the heading
21:21of quantum
21:23information
21:23processing
21:26or quantum
21:27information
21:27technologies
21:28and if I say
21:28information
21:29technologies
21:30to you
21:30you'd think
21:30computers
21:31probably.
21:32And for
21:33me
21:33I think
21:34really
21:34the real
21:35benefits
21:36the starkest
21:37benefit
21:38we will see
21:38will come
21:39as these
21:40things start
21:40to converge.
21:42Horizon
21:42like IBM
21:43quantum
21:43is a quantum
21:44computing
21:44company
21:45it's right
21:46there in
21:46the name
21:47Horizon
21:47quantum
21:48computing
21:49but we
21:50have dark
21:51fibre
21:51we have
21:52quantum
21:53communications
21:54channels
21:54from our
21:55offices
21:55to the
21:56national
21:56university
21:57of
21:57Singapore
21:57so we're
21:58on Singapore's
21:59national
21:59quantum
21:59safe
22:00network
22:00but the
22:01reason we
22:02do this
22:02our motivation
22:03in taking
22:03part
22:04is not
22:06entirely
22:06for just
22:07being able
22:08to send
22:08secure
22:09communication
22:09from point
22:10A to
22:10point B
22:10but rather
22:11to think
22:12about the
22:12convergence
22:13of quantum
22:13communication
22:14and quantum
22:15computing
22:16because that
22:17allows us to
22:17start doing
22:18secure computing
22:19in ways that
22:20aren't possible
22:20before
22:21you can perfectly
22:22hide quantum
22:23computation
22:23so you can run
22:24on an untrusted
22:25device and not
22:26leak any
22:26information at
22:28all about the
22:29program you're
22:29running to the
22:30person operating
22:31that device
22:32and you can
22:33guarantee that
22:33the results that
22:34come back are
22:34correct or detect
22:36any tampering
22:37with them
22:37so there's a lot
22:38that comes from
22:39the convergence
22:40of these
22:40technologies
22:41I think
22:42that's where
22:42we are
22:43most interested
22:45I'm not sure
22:46if I understood
22:47that quickly
22:47but that was
22:47a bit of
22:48almost doing
22:49the job
22:49of what
22:50blockchain
22:50is heralded
22:51as doing
22:52or solving
22:53some of the
22:54same security
22:56and data
22:57problems as
22:57that
22:57so in the
22:59classical world
23:00there's a
23:00technology called
23:01homomorphic
23:02encryption
23:02which is a
23:03way of
23:04encrypting data
23:05such that you
23:05can still
23:06compute on the
23:06data without
23:07revealing what
23:08the data is
23:09but there's
23:10quite high
23:10overhead for
23:11this and it's
23:12only secure
23:13under certain
23:13assumptions that
23:14some problems
23:15are computationally
23:17difficult and to
23:18be honest we
23:18don't really know
23:19how well any of
23:20these assumptions
23:20hold up in a
23:21world with large
23:22scale quantum
23:23computers
23:23on quantum
23:24computers you
23:25can do this
23:26kind of secure
23:26computing with
23:27extremely low
23:28overhead in a
23:29way that's
23:30provably secure
23:31where there are
23:31no assumptions
23:32made other than
23:33the laws of
23:33physics
23:35okay thank you
23:37I hope to come
23:38back to
23:39vivitech in
23:39five ten
23:40years and the
23:41halls will be
23:41full of
23:42quantum companies
23:43but at the
23:44moment they're
23:45not are they
23:46they're everything
23:47is artificial
23:48intelligence that
23:49this whole
23:50building is a
23:51wash with AI
23:52companies startups
23:53brands I want to
23:56end before I
23:56open the floor to
23:57questions to get
23:58your questions ready
23:58on the intersection
24:00of AI and
24:01quantum and we
24:02haven't spoken
24:02about this yet
24:08who's leading
24:09who is AI
24:09going to lead
24:10quantum is
24:10quantum going to
24:11lead AI how do
24:12these two
24:12technologies work
24:14together independently
24:17could one exist
24:18without the other
24:19Joe bros
24:21sure let me lead
24:23off so at the
24:25intersection of
24:26both quantum and
24:27AI there's AI for
24:30quantum and then
24:32and then there's
24:33quantum for AI
24:34let me take the
24:35first one AI for
24:38quantum a couple
24:39of application areas
24:41of artificial
24:42intelligence in
24:43quantum is in
24:44code assist so
24:46in what sorry
24:47this would encode
24:48assistance so
24:50this would be using
24:51artificial intelligence
24:52to write quantum
24:54algorithms as a
24:56code assistant
24:57another area is to
25:01learn the noise that
25:04exists in a quantum
25:05circuit and then
25:07help to identify the
25:10sources of that
25:11noise and debit them
25:12or subtract them out
25:14of the quantum
25:15circuit those are two
25:16applications of
25:18artificial intelligence
25:19in quantum computing
25:21now to go the other
25:22way for how quantum
25:26plays into AI a good
25:30example is in quantum
25:31machine learning so we
25:34have seen that you can
25:36create quantum
25:37kernels which are ways
25:40to take features from
25:42data and project them
25:44into higher dimensional
25:45spaces where you can do
25:46data segmentation using
25:49quantum kernels to
25:50develop those what are
25:53called gram matrices or
25:55those those projectors
25:57that take the data that
25:59appears inseparable
26:00putting it into a higher
26:02dimension using a
26:04quantum kernel trick
26:05that is something that is
26:08important to artificial
26:09intelligence and into
26:11machine learning so
26:13that's one way that
26:14quantum is playing into
26:16AI is in data
26:18segmentation in support
26:21vector machines for
26:22for AI okay wow thank
26:26you a lot of a lot
26:29yeah another dimension
26:30yes moving things
26:31wow
26:32Araceli
26:34yeah I'm gonna I'm gonna
26:36take a little bit a look
26:37on the implications for
26:39society if that's okay
26:40because I have been
26:41involved in thinking
26:42about really what it
26:44does it mean for the
26:45good and for the bad
26:46since 2018 we were I
26:48think five people are on
26:49the world thinking about
26:50this and then over the
26:51years there are more and
26:52more people we have been
26:53doing a lot of work with
26:54the World Economic Forum
26:55and you can read all
26:56these you can google it
26:58and you see more
26:59information what are
27:00really the implications
27:01of these new technologies
27:02for the good and for the
27:03bad and a lot of things
27:05it's like well quantum
27:06computing is kind of the
27:08same that AI right all
27:10the implications that we
27:10know that AI can be bad
27:12for it's gonna be the
27:14same quantum computing
27:15but we go again quantum
27:17information technologies is
27:19not just the software part
27:21that say that is kind of
27:22similar to AI there are
27:24other technologies that
27:26could have this dual use
27:27for example the quantum
27:29sensor that can be used to
27:31map the magnetic field or
27:33the gravity meter is the
27:35same technology that you
27:35could use to actually
27:37detect your enemy in a
27:39warfare so yes there is a
27:41lot of potential for the
27:43bad as well so the social
27:44implications of all these
27:45the ethical implications are
27:47very important to take into
27:48account what are the same
27:49things with AI and now the
27:51good thing is because of
27:52it's just here AI JGPT and
27:54all that we are learning so
27:56fast that hopefully we can
27:58learn when quantum is here
27:59and we are well prepared in
28:00advance that is not
28:01happening like now that I
28:03remember giving a training
28:04to lawyers and they were
28:05like we don't have
28:07regulations for AI and
28:08JGPT some people are
28:10coming with problems and we
28:11don't know what to do about
28:11it so the important thing
28:13about these destructive
28:14technologies in this case
28:15for quantum is get well in
28:17advance prepared what could
28:18be the implications even if
28:19we don't know a lot so
28:21that we are prepared for
28:23the good and for the bad.
28:25Awesome answer.
28:27So I think Joe did a great
28:29job of describing some of
28:31the applications of AI in
28:33quantum computing for
28:34characterising systems for
28:37code assist for copilot type
28:39stuff but if you go beyond
28:42this there's also things like
28:43designing error correction
28:44codes and decoders for error
28:46correction codes that have
28:47interesting potential.
28:50It's also true that if you
28:52look at it the other way
28:53around quantum computers have
28:55a lot of potential to assist
28:58AI or to accelerate AI.
29:01In particular there's two
29:04different kinds of artificial
29:06intelligence or quantum
29:07machine learning techniques that
29:09are explored.
29:10There's variational techniques
29:11which treat quantum programs a
29:14little bit like neural networks
29:15and try to train them with
29:17exactly the same techniques we
29:18use to train neural networks.
29:20And they're possible now.
29:22People run these.
29:23They run them on IBM systems.
29:25I think you have a library for
29:25it.
29:26They run them on many different
29:28systems and you can get
29:29reasonable results out of them.
29:30But it's very difficult to prove
29:32if there's an advantage for them.
29:34On the other hand, we know that you
29:36can accelerate certain quantum
29:39certain classical machine
29:41learning models.
29:43You can do the computations
29:44unallowing them exponentially
29:46faster on a quantum computer.
29:48And actually Horizon was founded
29:51coming out of a project that looked
29:53to automatically accelerate
29:54machine learning models.
29:55And there's a lot of potential there.
29:59I'd worked on Gaussian process
30:00regression myself as an academic
30:03before I started Horizon.
30:04But since then, we've seen
30:06acceleration for support
30:07vector machines.
30:08Actually, that predates our work.
30:10But also more recently for
30:11transformers, which underlie
30:13large language models and many
30:14of the recent advances in Gen AI.
30:17However, what I would say is
30:18that to get to these places,
30:20what we need is more than just
30:22the processor.
30:23We also need what's called
30:24quantum random access memory.
30:26And that's a kind of quantum version
30:28of the memory in your computer
30:30that allows the processor
30:32to read the memory
30:33in quantum superposition.
30:35Now, we don't have that yet.
30:38You know, there are good ideas
30:39on how to build it.
30:40But if we can get to that point,
30:42we'll see an enormous advantage
30:44for some machine learning models.
30:46But it's a little bit further off.
30:48It's not going to be
30:49the first applications we see.
30:51Okay, thank you.
30:52There you go.
30:53I promise I'd open the floor
30:54to any questions.
30:55So if anybody has a question
30:56for our panelists,
30:57please raise your hand
30:58and somebody will bring
30:59a microphone to you.
31:02Straight away.
31:03There you go.
31:03Thank you.
31:04Sir.
31:07Thank you.
31:09So just before that,
31:10we had a session with Mark
31:12and it was with an investor as well.
31:15So my question is,
31:18when it comes to,
31:19from the investment point of view,
31:21when it comes to bringing
31:23these use cases
31:24associated with quantum
31:26into life,
31:27how intensive are these
31:29from investment point of view
31:31to bring them to life?
31:33So I know every business model varies,
31:35but if you could give a sense
31:37from your business perspective
31:41regarding how capital intensive
31:44or how much investment
31:47is needed
31:48to bring these use cases to life,
31:50it will be great.
31:51Thank you.
31:52Did you get that?
31:53How much investment is needed
31:54to bring these use cases to life?
31:56And perhaps,
31:57Adedrin,
31:57are we there?
32:00Sure.
32:01So it depends.
32:02we don't really have
32:04useful quantum computers yet.
32:05We're not quite there.
32:07And there's going to be
32:07a lot of investment
32:08from the hardware companies
32:09over the next couple of years
32:11in trying to get there.
32:12Fortunately,
32:13there's been some really big advances
32:15in error correction
32:16that maybe moved the milestones
32:19closer than we thought
32:20they were before,
32:21which means we might be able
32:23to get to useful quantum computing
32:24with fewer qubits
32:26with smaller scale systems
32:27than we initially thought.
32:29We're also seeing
32:30some interesting developments
32:31in areas,
32:33sorry,
32:33to talk about neutral atoms
32:35with a superconducting company
32:37on the panel.
32:38But we're seeing
32:39really interesting advances
32:41with neutral atom quantum computing
32:43as well
32:43that is also moving things closer.
32:45So the hardware companies
32:46are definitely going to spend
32:47some money
32:48and that's going to be
32:49a large amount
32:50over the next couple of years.
32:51But if you compare that
32:53to what it cost
32:53to develop the iPhone,
32:55for example,
32:55it's still, you know,
32:58not all that big.
33:00And if you think about it
33:01that way,
33:02the upside is tremendous
33:04potentially compared
33:05to the iPhone.
33:06Actually,
33:07if this is computing
33:08all over again,
33:09if you go to our world
33:11and data
33:11and you look at
33:12the total economic output
33:13of the world
33:13over the last 150 years,
33:15there is a kink
33:16in that graph
33:17right where the first
33:18commercial computer
33:19had sold,
33:20and the world economy
33:21grows at double the rate
33:22it previously did
33:23from that point on
33:23to the present day.
33:25So the upside's
33:26tremendously high
33:27and definitely,
33:28in my view,
33:29justifies the investment.
33:31Can I say something?
33:32I just want to say
33:34something that I heard
33:35from an investor
33:36and also someone
33:37who works in a quantum company.
33:39Even if you put
33:40all the money
33:41of the world,
33:41you cannot accelerate
33:42science.
33:43And I love that.
33:45But it helps.
33:46And I think
33:47if you really look
33:48at what is happening
33:49in the quantum ecosystem,
33:50it really helps a lot.
33:52Love that.
33:53Brilliant.
33:53Thank you.
33:54Another question?
33:55Anybody?
33:56Over there?
33:58The lady on the third row?
34:07Thank you.
34:08I'll pick up
34:09from the parallelism
34:11between AI and quantum.
34:13Do you have any insights
34:15on the environmental impact
34:17of quantum,
34:18of the development
34:20of quantum technologies?
34:21because AI
34:22is very environmentally draining.
34:26So,
34:26do you have any thoughts
34:27on that?
34:28Thank you.
34:29Did you hear that?
34:30Any thoughts
34:30on the energy usage
34:33of quantum?
34:34Yes.
34:36Yes.
34:37We're targeting
34:38low energy usage.
34:40So,
34:41when you look
34:42at a quantum computer
34:43and you look
34:44at the amount
34:44of energy
34:45per solution,
34:47it's going
34:47to be significantly
34:48lower than
34:49high performance
34:50computing.
34:50for our large
34:52system 2,
34:54we are looking
34:55to do it
34:56at something
34:56around,
34:57for our largest
34:58system we have
34:59planned,
34:59at around 2 megawatts
35:01for the entire system
35:03and that compares
35:05to say 60
35:06or 70 megawatts
35:08in a very large
35:09high performance
35:10computer.
35:11So,
35:11bottom line,
35:13quantum computing,
35:14our target
35:15is to make it
35:16a very energy
35:17efficient
35:18type of computation.
35:21Thank you.
35:22And with that,
35:22we're out of time.
35:23Again,
35:23if you'd like
35:24to find our
35:24guests,
35:25panellists,
35:26harangue them
35:26after the show,
35:27ask them your
35:28questions,
35:29and a big round
35:30of applause
35:31for our panellists,
35:32please.
35:32Thank you.
35:33help you.
Commentaires