Latest / What's Your Problem? / How Quantum Computers Could Change the World

Transcript

How Quantum Computers Could Change the World · What's Your Problem?

0:15
Speaker 1: Pushkin. I'm Jacob Goldstein, and this is What's Your Problem,
0:34
Speaker 1: a show about people using technology to solve problems that matter.
0:38
Speaker 1: My guest today is Ben Bloom. He's the co founder
0:42
Speaker 1: and CEO of a company called Atom Computing, and his
0:45
Speaker 1: problem is this, how do you build a useful quantum computer?
0:51
Speaker 1: If Ben succeeds, or if one of the other companies
0:54
Speaker 1: working on quantum computers succeeds, quantum computers could make profound
0:59
Speaker 1: improvements in everything from discovering new medicines to building cheaper
1:03
Speaker 1: ways to store energy. Quantum computing today kind of reminds
1:07
Speaker 1: me of where AI was, say, ten or fifteen years ago.
1:11
Speaker 1: Huge possibility, lots of people working on it. Still not mainstream,
1:16
Speaker 1: but it's worth talking about now for a few reasons.
1:19
Speaker 1: For one thing, if or when quantum computers do work,
1:23
Speaker 1: they will be an extremely big deal. The science is
1:27
Speaker 1: clear that they can solve problems that are just too
1:30
Speaker 1: complex for traditional computers or even AI to ever solve.
1:35
Speaker 1: There are the potential energy and medical breakthroughs I mentioned before. Also,
1:40
Speaker 1: quantum computers can crack a common, widely used form of encryption.
1:44
Speaker 1: At the moment, giant tech companies like Google and IBM
1:48
Speaker 1: and Amazon are spending billions of dollars on their own
1:51
Speaker 1: quantum computing projects. Several smaller companies, including Ben Bloom's Atom Computing,
1:57
Speaker 1: have had money pour in from venture capitalists and public markets,
2:01
Speaker 1: and the Chinese government is spending billions more on its
2:04
Speaker 1: own quantum computing project. A lot of money and a
2:08
Speaker 1: lot of smart peace people and a high stakes outcome
2:10
Speaker 1: if it works. That's why quantum computing is worth talking about. Now,
2:15
Speaker 1: before we get to the interview, here's the basic idea
2:17
Speaker 1: of how a quantum computer is different than a traditional computer,
2:20
Speaker 1: a classical computer. In a classical computer, as you probably know,
2:24
Speaker 1: the basic unit of information is a bit. A bit
2:28
Speaker 1: can only be one of two.
2:29
Speaker 2: Things, zero or one off or on. Amazingly, everything that
2:36
Speaker 2: computers do is just built on lots and lots and
2:39
Speaker 2: lots of zeros in ones. And you can do a
2:42
Speaker 2: lot with zeros and ones. But you cannot do everything.
2:45
Speaker 2: For one thing, you can't crack standard methods of online encryption.
2:50
Speaker 2: For another, you can't build a complete model of even
2:54
Speaker 2: a simple molecule, say the kind of molecule you would
2:58
Speaker 2: use as a drug. This is where quantum computers come
3:01
Speaker 2: in quantum computers are not built out of bits, out
3:04
Speaker 2: of zeros and ones. They're built out of what are
3:07
Speaker 2: called cubitsum bits. To build a quantum bit, you use
3:12
Speaker 2: a quantum particle.
3:14
Speaker 3: Then.
3:14
Speaker 1: Bloom's company is called Atom Computing because they build each
3:18
Speaker 1: cubit with a single atom, and when the quantum computer
3:21
Speaker 1: is working, each atom, each cubit is not limited to
3:25
Speaker 1: being in a single state, to being just a one
3:27
Speaker 1: or zero in a weird quantumy way, a single cubit
3:32
Speaker 1: can be in many different states at once. And on
3:35
Speaker 1: top of that, when you combine cubits, what happens to
3:39
Speaker 1: each one instantly affects the others. What this means in
3:43
Speaker 1: practice is that a quantum computer with enough cubits could
3:46
Speaker 1: solve some problems that the most powerful classical computers literally
3:51
Speaker 1: could not solve in a million years. But quantum computers
3:56
Speaker 1: today are too small to do that, or at least
3:59
Speaker 1: they can't do it for any practical worldly problems. They
4:02
Speaker 1: just don't have enough cubits, and as Ben and I
4:05
Speaker 1: discuss in our conversation, building a function in quantum computer
4:09
Speaker 1: with lots of cubas is in fact a very hard
4:11
Speaker 1: engineering problem. So to start, I asked Ben to tell
4:15
Speaker 1: me how the field of quantum computing will look in
4:17
Speaker 1: a few years, when, if things go well, those engineering
4:20
Speaker 1: problems will have been solved.
4:23
Speaker 3: So I think in five years what you'll see is
4:28
Speaker 3: quantum computers attacking problems for national labs, for defense, for
4:35
Speaker 3: all these kind of esoteric industries who could call them
4:39
Speaker 3: that want calculations done that cannot be done on classical computers.
4:44
Speaker 3: It's probably going to be something like materials simulations. So
4:47
Speaker 3: how electrons behave in materials?
4:49
Speaker 4: And why are national labs interested in that? Ooh?
4:54
Speaker 3: I think that's a funny question. I think the Department
4:57
Speaker 3: of Energy is really interested in understanding materials at their extremes.
5:03
Speaker 4: OK.
5:04
Speaker 3: And I think that's true to both make it more
5:07
Speaker 3: efficient for the USA government to deal with all of
5:11
Speaker 3: the equipment they have, and also really true I mean,
5:15
Speaker 3: obviously for the Department of Energy main mission, which is
5:19
Speaker 3: understanding nuclear weapons and doing that without having to test
5:22
Speaker 3: nuclear weapons.
5:23
Speaker 1: So five years from now, quantum computers are where classical
5:27
Speaker 1: computers were in the nineteen forties, right, A few giant,
5:31
Speaker 1: crazy expensive ones being used by the government or a
5:35
Speaker 1: couple governments. Two things people talk a lot about with
5:38
Speaker 1: quantum computing are drug discovery, discovery of new drugs, and energy.
5:43
Speaker 1: For some reason, people talk about energy, they talk about batteries,
5:46
Speaker 1: that kind of thing. So if things go relatively well
5:49
Speaker 1: in quantum computing, like, are we going to be seeing
5:51
Speaker 1: meaningful impacts on those fields and what in ten years
5:54
Speaker 1: and if so, how will they be different?
5:58
Speaker 3: Yeah? I do think we will. I mean, I think
5:59
Speaker 3: ten fifteen years we will see new drugs on the
6:02
Speaker 3: market that we're designed on a quantic computer. I think
6:05
Speaker 3: we will see batteries they can do more recharge cycles
6:09
Speaker 3: because we understand the materials and understand the various properties
6:12
Speaker 3: that you know, cause them to lose charge over recharge cycles, and.
6:16
Speaker 1: More recharge cycles means cheaper in the long run importantly, right,
6:20
Speaker 1: Like batteries are constrained by cost now, and if one
6:23
Speaker 1: battery can last longer, that effectively means you it's cheaper
6:27
Speaker 1: per cycle.
6:28
Speaker 4: Yeah.
6:28
Speaker 3: And I think another example of this is, you know,
6:30
Speaker 3: the one that people throw out a lot is kind
6:33
Speaker 3: of understanding, you know, how you actually do fertilizer production. Like,
6:37
Speaker 3: it turns out a few percent of the world's energy
6:40
Speaker 3: is spent on fertilizer production. And so if you can
6:42
Speaker 3: find a new catalyst that could do that same process
6:47
Speaker 3: with slightly lower temperature or slightly less pressure, you know,
6:52
Speaker 3: marginal savings in that process turn into gigantic global savings
6:58
Speaker 3: in an energy budget sense. And so all of these
7:00
Speaker 3: problems kind of have this general idea behind them, which
7:03
Speaker 3: is that chemicals and materials and understanding the physical processes
7:07
Speaker 3: which we kind of have built our old on top
7:09
Speaker 3: of understanding them to a point where you could engineer
7:13
Speaker 3: around them or engineer them to work better, could have
7:17
Speaker 3: these you know, giant lever arm effects.
7:19
Speaker 1: So it's basically, I mean a more and better drugs
7:23
Speaker 1: and then be just efficiency gains. So that's fewer emissions,
7:27
Speaker 1: more power, lower costs. Like those are the dreams broadly.
7:31
Speaker 4: Yeah, okay, so where are we today?
7:36
Speaker 3: Yeah, we're building toy systems. I mean, there's no question
7:41
Speaker 3: I think that over the past I would say two years,
7:45
Speaker 3: a lot of technologies, including the one I work on,
7:47
Speaker 3: neutral atoms, have kind of crossed a threshold where you
7:50
Speaker 3: can build better and better cubits by taking that quantum information,
7:55
Speaker 3: spreading it around and doing error correction. And one of
7:58
Speaker 3: the examples that you know is probably what some sense
8:01
Speaker 3: started this entire race is it turns out you can
8:04
Speaker 3: factor numbers on a quantum computer efficiently.
8:06
Speaker 1: Which sounds trivial to the initiated, but it turns out,
8:10
Speaker 1: among other things, if I have this right to allow
8:13
Speaker 1: you to crack most of the encryption on the Internet,
8:16
Speaker 1: true or not true, It's true.
8:18
Speaker 3: The kind of encryption we have now was was a choice,
8:21
Speaker 3: and in some sense right now it's even a choice
8:23
Speaker 3: in your browser or in your operating system. And the
8:27
Speaker 3: likes of you know, Apple or you know, the Linux
8:31
Speaker 3: maintainers or Microsoft, they can literally just you know, quote
8:35
Speaker 3: unquote flip a switch and it can switch to what
8:38
Speaker 3: they call is post quantum cryptography.
8:39
Speaker 1: So people have come up with new systems that are
8:42
Speaker 1: quantum resistant that can work to secure our data in
8:47
Speaker 1: the post quantum world.
8:48
Speaker 3: Exactly. Yeah, and so I think that the future is
8:51
Speaker 3: bright in that sense. Now. The funny thing that you know,
8:53
Speaker 3: governments around the world are a little probably freaked out
8:56
Speaker 3: about is the past, which is that we've been using
9:00
Speaker 3: for you know, decades upon decades, we've been transmitting information
9:05
Speaker 3: between you know, people and businesses and government enter prizes
9:09
Speaker 3: and stuff like that. That if someone stored it and
9:13
Speaker 3: waited until they had a quantum computer. They can just
9:16
Speaker 3: go and read in the future when quantic computers are available.
9:19
Speaker 3: And so I think that is the kind of issue
9:22
Speaker 3: that I think most people think of when they think, oh,
9:24
Speaker 3: they're going to break encryption. How bad is that going
9:26
Speaker 3: to be. It's not going to be, you know, five
9:28
Speaker 3: years from now that people are worried about how do
9:30
Speaker 3: you pay for something online? It's more going to be
9:33
Speaker 3: these kind of you know, big government issues of Okay,
9:37
Speaker 3: you know, twenty years ago we sent a cable to
9:39
Speaker 3: so and so and it said this.
9:41
Speaker 1: I mean, we can let's talk about geopolitics here. Now,
9:44
Speaker 1: when people say let's talk about geopolitics in the US,
9:46
Speaker 1: they basically mean let's talk about China, and specifically, I.
9:50
Speaker 4: Mean, what happens if China gets there first?
9:52
Speaker 1: Right? Like, I do feel like, I know that's a
9:54
Speaker 1: crude way to ask the question, but I also know
9:56
Speaker 1: China is putting a lot of money into quantum computing,
9:58
Speaker 1: right Like? Is there something of a race here? Is
10:01
Speaker 1: that a way to think about it?
10:03
Speaker 3: I think it is a race. I mean, I think
10:04
Speaker 3: that they're doing it differently than the US. I mean,
10:08
Speaker 3: at the US. You know, my company out in Computing
10:10
Speaker 3: is one of many companies in the United States who
10:13
Speaker 3: are pursuing quantic computing, are trying to build a quantic computer.
10:18
Speaker 3: I think it's all private. You know, there is some
10:20
Speaker 3: public funding which we're accessing. There's Darper programs and things
10:23
Speaker 3: like that, but it's mostly private funding. Whereas you talk
10:27
Speaker 3: about what's going on in China, and I think there
10:29
Speaker 3: are big governmental initiatives, and I also think there are
10:33
Speaker 3: in some sense state backed companies that exist.
10:36
Speaker 1: This is very much in keeping with both of our countries, right,
10:40
Speaker 1: exactly what you would expect.
10:42
Speaker 4: Perhaps it's a test of our various systems.
10:45
Speaker 3: Yeah, yeah, So, I mean I think that the question
10:47
Speaker 3: of you know, is it bad if China, you know,
10:51
Speaker 3: builds a quantum computer and we don't have one, I mean, potentially,
10:53
Speaker 3: I mean, I think that a lot of the advances
10:56
Speaker 3: we were talking about earlier. Yeah, there are performance gains,
11:00
Speaker 3: efficiency gains and things like that, but those are the
11:02
Speaker 3: kinds of things that really I would say, drive an
11:04
Speaker 3: economy in the sense where the numbers associated with them
11:08
Speaker 3: are so so large that it's hard for me to
11:12
Speaker 3: wrap my head around. Like if you can decrease the
11:14
Speaker 3: energy budget of your country by five percent. Uh, that's
11:19
Speaker 3: a huge number, and I think it can actually make
11:22
Speaker 3: big changes possible.
11:23
Speaker 4: Yeah. I mean in a way, it makes you much richer.
11:25
Speaker 1: Right, it allows you to do things you could otherwise
11:28
Speaker 1: not afford to do, or you wouldn't otherwise have the
11:30
Speaker 1: resources to do. Y. I mean, presumably it gives you
11:36
Speaker 1: some kind of cryptographic power.
11:40
Speaker 3: Well, I mean it goes back to this idea that
11:42
Speaker 3: there's you know, there's this store now, decrypt later philosophy,
11:47
Speaker 3: which I've heard, you know, espoused by various government officials
11:50
Speaker 3: as you know, others are doing this and so on
11:53
Speaker 3: and so forth, which is that they're you know, storing
11:55
Speaker 3: all the communications and things like that. I think that
11:59
Speaker 3: is a problem. There's no question if if you know,
12:01
Speaker 3: we're successful and we build quantic computers, you will be
12:04
Speaker 3: able to decrypt later. I guess the question is how
12:08
Speaker 3: useful is old information? That would be the problem, And
12:11
Speaker 3: the question is how how bad is that going to be?
12:14
Speaker 3: Maybe it's not so so bad on individual scale, but
12:17
Speaker 3: on a kind of company scale or a country scale,
12:20
Speaker 3: maybe it could be quite quite problematic.
12:23
Speaker 1: So let's talk a little more about the difference between
12:26
Speaker 1: quantum computers and regular computers, classical computers. And maybe one
12:31
Speaker 1: way to do that is to talk about classical computers,
12:33
Speaker 1: the kind of computers we have now, the kind of
12:35
Speaker 1: computers that underpin AI are bad at, like what is
12:39
Speaker 1: their meaningful limitation in this context.
12:42
Speaker 3: So, say you have an electron and you want to
12:45
Speaker 3: understand the properties of that electron. And because you're designing
12:48
Speaker 3: a drug, you're you're looking at a material, You're doing
12:51
Speaker 3: something like that.
12:52
Speaker 1: A particular electron on a particular molecule.
12:55
Speaker 3: Yah, yeah, yeah, yeah, And all you want your classical
12:58
Speaker 3: computer to do is write down the state of that electron,
13:01
Speaker 3: very very precisely, where is it and what is it doing,
13:05
Speaker 3: and it's spin and things like that. Yeah yeah, Now wow.
13:09
Speaker 3: You try to write that down, and it turns out
13:11
Speaker 3: you use a lot of binary data. You use a
13:14
Speaker 3: lot of classical data to do that. Now, if you
13:16
Speaker 3: want two electrons, you have to use more classical data
13:21
Speaker 3: describe it. But it turns out, and this is where
13:24
Speaker 3: quantum mechanics starts mattering. It turns out you don't just
13:28
Speaker 3: care about the individual electrons in their states. You actually
13:31
Speaker 3: care about all the correlations, all the ways the two
13:34
Speaker 3: electrons are interacting with each other and are linked together,
13:38
Speaker 3: and that turns out to require more classical data to
13:41
Speaker 3: write down and very very quickly with tens to less
13:45
Speaker 3: than one hundred electrons. If you tried to write down
13:48
Speaker 3: all the classical data you needed to describe that situation
13:53
Speaker 3: you would need, I think it was more bits of
13:57
Speaker 3: information than there are, like atoms in our universe.
14:01
Speaker 1: It would be impossible. It would just be impossible for
14:03
Speaker 1: a classical computer yourself. And so it is the case
14:06
Speaker 1: that like once you get to the really fundamental level
14:12
Speaker 1: of what's going on with energy or what's going on
14:14
Speaker 1: with matter, with materials, it behaves in a quantum way,
14:19
Speaker 1: and classical computers can't understand what's going on once you
14:23
Speaker 1: get to that level, and quantum computers, at least the
14:27
Speaker 1: theoretical quantum computer, if you could solve the engineering problems,
14:31
Speaker 1: could do it.
14:32
Speaker 3: Yeah exactly. I mean, I think that the cool thing
14:37
Speaker 3: about quantum computing we kind of do have Norse stars.
14:41
Speaker 3: We know that if you can build a quantic computer,
14:43
Speaker 3: you can factor large numforts. We know that if you
14:46
Speaker 3: can build a quantic computer that you know, can deal
14:49
Speaker 3: with billions of operations that you can find the you know,
14:52
Speaker 3: structure of some molecule or something like that. And so
14:55
Speaker 3: we know that the goal is go from the short
14:58
Speaker 3: little calculations you can do now to longer and longer
15:02
Speaker 3: and longer calculations to the point where you can reach
15:05
Speaker 3: these goals.
15:05
Speaker 4: Yes, and so maybe there's one.
15:09
Speaker 1: There's one sort of problem broadly stated that seems particularly
15:17
Speaker 1: interesting and worth discussing.
15:21
Speaker 4: Let me see if I have this right? Is it
15:23
Speaker 4: right that.
15:26
Speaker 1: Each bit, each cubit of a quantum computer needs to
15:30
Speaker 1: be entirely isolated. It needs to be sort of not
15:34
Speaker 1: in touch with anything in the world. No, not a photon,
15:38
Speaker 1: not another molecule, not anything, right, because it is sort
15:41
Speaker 1: of sitting in this what a quantum superposition I'm reaching here,
15:47
Speaker 1: and if anything, if anybody sees it, if any light
15:50
Speaker 1: photon hits it, if anything happens to it, it kind
15:52
Speaker 1: of breaks, it decoheres and it doesn't work right? And
15:56
Speaker 1: is it fair that it's extremely hard for maybe obvious reasons,
16:00
Speaker 1: everything is touching everything else all the time. Is that
16:03
Speaker 1: like a central problem? Is that a problem worth talking about?
16:07
Speaker 3: It is because I think it's actually that the fundamental
16:11
Speaker 3: challenge of building a quantum computer is having that isolation,
16:15
Speaker 3: but then making sure that you have complete control over
16:18
Speaker 3: that quantum system.
16:19
Speaker 4: Right, And like, how could you have both at once?
16:22
Speaker 3: Right?
16:22
Speaker 1: If it's completely isolated, how can you control it? So,
16:26
Speaker 1: I mean, let's talk a little bit about your approach.
16:30
Speaker 1: You are working with Microsoft to build a thing for
16:36
Speaker 1: the Novo NORDESK foundation and what is part of the
16:39
Speaker 1: Danish government?
16:41
Speaker 4: Yeah? Yeah, so what are you building for the Danes? Yeah?
16:45
Speaker 3: So it's a twelve hundred cubit system, okay, And that
16:50
Speaker 3: is a kind of our part of this, which is
16:52
Speaker 3: that we're building the hardware. So it's a system that
16:55
Speaker 3: has all of the classical control around it, all the
16:58
Speaker 3: quantum control around it, so that they can trap, cool,
17:02
Speaker 3: manipulate twelve hundred physical cubits.
17:04
Speaker 4: Okay, and each one of those is one atom.
17:07
Speaker 3: Yes, each one of those is one atom.
17:08
Speaker 4: Yeah, very small number of atoms.
17:10
Speaker 3: Yeah. And actually the quantum part of our system is
17:13
Speaker 3: very very small, Like it's like half a millimeter by
17:16
Speaker 3: half a millimeter. It's very very.
17:17
Speaker 4: There's like a box or something. What is it? What
17:20
Speaker 4: is it?
17:20
Speaker 3: Yeah? Yeah, No, it is pretty much.
17:22
Speaker 1: A box half a millimeter by half a millimeter, which
17:26
Speaker 1: is I don't know what, like a size of like
17:29
Speaker 1: a hair or something like.
17:30
Speaker 4: It's so small.
17:31
Speaker 3: It's a little bigger than it.
17:32
Speaker 4: Okay, but yeah, what is it like? But it's bigger
17:34
Speaker 4: than like a little fingernail trimming.
17:36
Speaker 3: Right, yeah, fingernail trimming.
17:37
Speaker 1: That's a good bigernail trimming. And that's where all of
17:41
Speaker 1: the action is. That's the that's it. That's all of
17:43
Speaker 1: your Yeah, I guess twelve hundred atoms. Yeah, And what's
17:48
Speaker 1: going on in that tiny box? What do you start with?
17:52
Speaker 1: It's a metal, right, what's the element?
17:54
Speaker 3: Uh? Uturbium?
17:55
Speaker 1: Okay, so you start with this element, just a thing
17:57
Speaker 1: that exists in the world called U turbium. You don't
18:00
Speaker 1: need very much of it.
18:01
Speaker 4: What do you do?
18:03
Speaker 1: So?
18:03
Speaker 3: First off, we have to take we have a chunk
18:05
Speaker 3: of it, so we have to heat it up. So
18:08
Speaker 3: you go from a solid gas. That gas kind of
18:11
Speaker 3: streams down our system. We have lots of lasers kind
18:14
Speaker 3: of hit that gas so that it goes from hundreds
18:17
Speaker 3: of kelvin down to about a microkelvin, so ten to
18:21
Speaker 3: the minus six kelvin, so very very cold.
18:24
Speaker 1: And zero zero degrees kelvin is absolute zero. It's as
18:27
Speaker 1: cold as anything in the universe can ever be. So
18:30
Speaker 1: you're making it very, very very cold. And is that
18:33
Speaker 1: because you don't want it moving around, because you want
18:35
Speaker 1: it isolated?
18:36
Speaker 3: That's the exactly okay, yeah, pretty much like the temperature
18:39
Speaker 3: is pretty much the velocity it goes at and things
18:41
Speaker 3: like that. So we get the atoms, we get them
18:44
Speaker 3: down to a microkelvin, and then we kind of shuttle
18:48
Speaker 3: them using a laser. We kind of push them up
18:50
Speaker 3: into that tiny little area that's half a millimeter by
18:53
Speaker 3: half a.
18:54
Speaker 4: Million little box that's computer sort of.
18:58
Speaker 3: And then all of a sudden we get to this
18:59
Speaker 3: little area. And what we do through that microscope objective
19:03
Speaker 3: is we put a big beam into the back of
19:05
Speaker 3: that microscope objective. And the reason why we put a
19:08
Speaker 3: big beam when I say, it's like twenty millimeters yeah, okay,
19:11
Speaker 3: So you're shooting light through the microscope lens, let's say
19:14
Speaker 3: lens lens, okay, and when you do that, you put
19:18
Speaker 3: a big light beam into the back of that lensky
19:22
Speaker 3: What it does is it focuses down really really tight,
19:24
Speaker 3: and it creates an optical tweezer. Okay, And this won
19:27
Speaker 3: the Nobel Prize, and I think it was like twenty
19:30
Speaker 3: nineteen or something like that. The cool thing here is
19:34
Speaker 3: that this optical tweezer at the very focus of that
19:37
Speaker 3: lens gets really really tight and diverges really really quickly
19:42
Speaker 3: and for some light, and we very very carefully choose
19:45
Speaker 3: what kind of laser we're doing this with. Atoms are
19:49
Speaker 3: tracted to the point of highest intensity, okay, and so
19:52
Speaker 3: the atom kind of gets sucked in to the optical
19:55
Speaker 3: tweezer and then it wants to sit at the exact
19:58
Speaker 3: center of that pholk.
20:00
Speaker 1: So it's called an optical tweezer because it is grabbing
20:02
Speaker 1: a single atom.
20:04
Speaker 3: Yeah, okay, So what we do is we do a
20:08
Speaker 3: bunch of tricks so that we we don't just create
20:09
Speaker 3: one optical tweezer, but generally speaking, we build display technology
20:16
Speaker 3: that creates the image of many, many optical tweezers and
20:20
Speaker 3: just stuffs that into the back of the lens.
20:23
Speaker 1: So you basically have like what twelve hundred optical tweezers
20:27
Speaker 1: in each way, he grabs one atom and exactly sticks
20:30
Speaker 1: it in the box.
20:31
Speaker 3: Exactly. No, no, that is exactly what we do.
20:33
Speaker 1: So now there's this box that's ready to help someone
20:37
Speaker 1: figure out something about the world. What is something about
20:40
Speaker 1: the world that it might actually help them understand.
20:44
Speaker 3: Yeah, So, say there's a user in Denmark and Copenhagen
20:47
Speaker 3: who wants to understand the ground state of some molecule.
20:50
Speaker 3: They write down an algorithm that says, I am going
20:54
Speaker 3: to simulate many, many electrons, and I am going to
20:58
Speaker 3: use it such that it kind of gets down to
21:01
Speaker 3: the lowest energy state, Okay, and then I will read
21:04
Speaker 3: out the cubits in their various positions and things like that. Aha,
21:08
Speaker 3: so that we can actually understand what was the state
21:10
Speaker 3: of the electrons in this simulated system.
21:13
Speaker 1: So are they're sort of saying to the quantum computer
21:17
Speaker 1: that you have built, act like you are this molecule.
21:20
Speaker 1: Act like you are the electrons in this molecule exactly
21:23
Speaker 1: in their lowest energy state, and tell me what state
21:26
Speaker 1: you're in, Yes, exactly. We'll be back in just a minute. Hey,
21:42
Speaker 1: it's Jacob, and I want to tell you that I
21:44
Speaker 1: am hosting a new show called Business History. It's about
21:47
Speaker 1: the incredible innovations and massive failures and unbelievable characters in
21:52
Speaker 1: the history of business. And I hope I think the
21:55
Speaker 1: show provides insights about how business works today. At the
21:59
Speaker 1: end of today's episode of What's Your Problem, We're going
22:01
Speaker 1: to play you a clip from Business History.
22:04
Speaker 4: It's the story.
22:04
Speaker 1: Behind the video game company Atari, and they're surprising early
22:08
Speaker 1: hire of a young hippie named Steve Jobs. The show's
22:11
Speaker 1: called Business History. You can listen to it wherever you're
22:14
Speaker 1: listening right now, and will play that clip at the end.
22:17
Speaker 4: Of today's episode.
22:22
Speaker 1: One of the big problems in quantum computing is detecting
22:25
Speaker 1: and correcting errors. This is a hard problem for a
22:28
Speaker 1: few reasons. One of those reasons gets at the heart
22:31
Speaker 1: of quantum weirdness, and it's this. When a quantum computer
22:35
Speaker 1: is working, the cubits are essentially in many states at once.
22:39
Speaker 1: But if at a given moment you will look at
22:42
Speaker 1: a cubit try and figure out what state it's in,
22:45
Speaker 1: it will instantly snap into a single state. All that
22:48
Speaker 1: beautiful quantum weirdness of multiple states at once will suddenly disappear,
22:52
Speaker 1: and your quantum computer will not work. There's been a
22:56
Speaker 1: lot of progress on this in the past few years,
22:59
Speaker 1: but for the most part, quantum error correction still only
23:02
Speaker 1: works at a scale too small to be useful for practical,
23:05
Speaker 1: real world problems. I asked Ben, why.
23:09
Speaker 3: All the classical stuff we're building, all the laser projectors,
23:12
Speaker 3: all the spot makers, all these things. As you scale
23:15
Speaker 3: those numbers up, if you know you're a stray light
23:19
Speaker 3: or whatever it is you're doing for your quantum computer,
23:21
Speaker 3: if that increases, then all of a sudden, your quantumeric
23:24
Speaker 3: corection doesn't quite work as well as you want it to.
23:29
Speaker 4: It seems very hard, yes, very hard.
23:32
Speaker 1: Like I mean, I talk to a lot of Italy
23:33
Speaker 1: do things that sae hard, but this seems like wild hard.
23:38
Speaker 3: I would say that the last twenty four months there
23:41
Speaker 3: have been so many amazing advances. I would say everyone
23:44
Speaker 3: in the industry is more excited than they've ever been
23:48
Speaker 3: because I think you go back two years ago and
23:51
Speaker 3: no one had demonstrated quantumeric correction at any scale. Now
23:55
Speaker 3: there's US, there's Google, there's lots of other companies that
23:59
Speaker 3: have now demonstrated quantumeric corection. Yeah, a small number of cubits,
24:03
Speaker 3: but I think that is a key milestone, which is
24:05
Speaker 3: to say, look that theory is right, Like your errors
24:09
Speaker 3: go down as you spread this quantum information across your
24:12
Speaker 3: quantum processor. And I think that's a huge, huge boon
24:15
Speaker 3: because it says, no, all you have to do is
24:18
Speaker 3: just get to larger and larger and larger numbers.
24:21
Speaker 1: So there's this phrase, I think I heard you use
24:24
Speaker 1: it or maybe I read it, which is commercial advantage. Right.
24:27
Speaker 1: You've talked about sort of governments being initial users. What's
24:33
Speaker 1: the universe where a quantum computer becomes practical for a
24:37
Speaker 1: private company?
24:38
Speaker 3: Yeah, I mean I think that like a pharmaceutical company
24:41
Speaker 3: like Novo Nordisk or Eli Lilly or something like that.
24:44
Speaker 3: I think they're going to have simulation frameworks that you know,
24:48
Speaker 3: engineers are going to sit down at their desks and
24:51
Speaker 3: use those simulation frameworks, and it's going to be constantly
24:53
Speaker 3: sending jobs to a quantum computer.
24:56
Speaker 4: Tell me about this molecule in its lowest.
24:59
Speaker 3: Energy state exactly, yeah, or dynamics or something like.
25:02
Speaker 1: That, like what are the dynamics if this drug binds
25:06
Speaker 1: with this part of a cell?
25:08
Speaker 3: Exactly? No, exactly. So I think that most companies will
25:13
Speaker 3: view quantic computers in that lens where it's just another
25:16
Speaker 3: cloud computing resource that they're spending money on. And I
25:19
Speaker 3: could even imagine a future where you know, this is
25:22
Speaker 3: so ingrained in kind of simulation frameworks. It's not even
25:26
Speaker 3: clear you know, you're spending it on a quantum computing
25:30
Speaker 3: you know, hours on a quantic computer. You're actually just
25:33
Speaker 3: you know, buying time on your simulation framework that you
25:35
Speaker 3: get from another company or so on.
25:37
Speaker 1: Yes, we haven't talked about AI. We've gone a while,
25:41
Speaker 1: we haven't talked about AI. And there's like a couple
25:43
Speaker 1: of sides of AI, right, There's like AI helping to
25:46
Speaker 1: make quantum computers. There's also AI being able to do things,
25:50
Speaker 1: you know, on classical chips that we didn't think maybe
25:53
Speaker 1: classical computers could do. Like I'm thinking of of the
25:56
Speaker 1: protein folding problem, right, this famous hard molecular level problem
26:02
Speaker 1: that AI solved that nobody could solve for a long time,
26:06
Speaker 1: seems like the kind of thing that you might have
26:08
Speaker 1: looked years ago been like, oh, here's the thing quantum
26:12
Speaker 1: computers can do. They can predict the shape of a protein.
26:14
Speaker 3: Yeah, I think that's true, I will say, when we
26:17
Speaker 3: go and talk to you know, the experts in the field.
26:20
Speaker 3: I think one of the things we hear an awful
26:21
Speaker 3: lot about is is the idea that the AI is
26:25
Speaker 3: only ever good at the data you give it and
26:28
Speaker 3: a new use case for quantum computers that you know,
26:30
Speaker 3: no one talked about more than like three years ago,
26:34
Speaker 3: was using quantum computers to just pump out data for
26:37
Speaker 3: a train, and.
26:40
Speaker 1: I it would be if that was the case, it's like, oh,
26:44
Speaker 1: we're just making the AI better. We kind of a
26:47
Speaker 1: bummer at sub level, not to be silly, but you
26:49
Speaker 1: know what I mean.
26:55
Speaker 3: But I do think it's this idea that you know, fundamentally,
26:57
Speaker 3: like if all of your AI data is coming from
27:01
Speaker 3: classical models or classical you know, algorithms spitting out data
27:06
Speaker 3: to go and train alpha fold or whatever it's called on,
27:10
Speaker 3: you only get it so good. And if you can
27:12
Speaker 3: go a step further and start getting the AI to
27:16
Speaker 3: understand the kind of quantum pieces of the puzzle, that
27:20
Speaker 3: all of a sudden your AI is going to start
27:21
Speaker 3: including those in its predictions. But I do agree that
27:26
Speaker 3: the flip side of this is building a ton of
27:29
Speaker 3: quantum systems whose sole job it is just pumped out
27:32
Speaker 3: data for FREYI is a little surprising.
27:34
Speaker 4: A little bit of a sad trombone.
27:36
Speaker 1: And I mean, is there any way in which AI
27:41
Speaker 1: is helping you or helping the field?
27:44
Speaker 4: Figure things out.
27:45
Speaker 3: Yeah, I mean I do think it is like I
27:47
Speaker 3: think that actually in control, I think they're a huge,
27:50
Speaker 3: huge ability for AI. Like I think that when we
27:53
Speaker 3: talk about scaling up systems and how do you tune
27:56
Speaker 3: things and how do you kind of get control to
27:59
Speaker 3: across you know, many many, many cubits, AI is much
28:03
Speaker 3: much better at finding correlations and data and things like
28:05
Speaker 3: that and actually having an understanding of how you turn
28:08
Speaker 3: the troll knobs you have into the outputs you want.
28:12
Speaker 3: AI is actually fantastic at that. I also think the
28:14
Speaker 3: other thing is true, which is just that I mean
28:17
Speaker 3: we operate faster because people write software faster now, oh right,
28:20
Speaker 3: Like even in the last like six months, I would
28:22
Speaker 3: say there is more and more and more software at
28:24
Speaker 3: Atom Computing that is being written with AI, and it
28:28
Speaker 3: would be done you know, I don't even know ten
28:31
Speaker 3: times faster with AI than it.
28:32
Speaker 1: Was possible speaking of simple productivity gains with profound cons Yeah.
28:40
Speaker 1: So well, we started out talking about a you know,
28:43
Speaker 1: relatively optimistic scenario five years, ten years, like why might
28:49
Speaker 1: quantum computing not get there or take much longer than
28:53
Speaker 1: you think?
28:55
Speaker 3: I think it's all a challenge of you know, how much.
28:58
Speaker 3: How many resources are you willing to devote to get
29:03
Speaker 3: quantum computers to work? And what are the use cases?
29:06
Speaker 3: Like I think that a lot of the kind of
29:08
Speaker 3: chemistry use cases and stuff we're excited about. But like
29:12
Speaker 3: you said, there's always going to be new techniques and
29:14
Speaker 3: things like that. And if we can find new techniques
29:18
Speaker 3: that you know, get us far enough, then the thing
29:20
Speaker 3: that we only have to fall back on is decryption.
29:24
Speaker 3: And the question is, you know who there's only one
29:26
Speaker 3: customer that wants to decrypt at least in the United.
29:29
Speaker 4: States, at least two. I was gonna say every.
29:31
Speaker 3: Government in the world, right, Yeah, But I think it's
29:35
Speaker 3: it's a question of like resources and resource allocation, which
29:39
Speaker 3: is that if that's something the government's you know, very
29:42
Speaker 3: very excited about still, then of course, you know, I
29:44
Speaker 3: think we can keep scaling and we can make bigger
29:46
Speaker 3: and bigger quantic computers.
29:47
Speaker 4: And this is the use case you say, it's just money.
29:50
Speaker 3: I mean I think that we have and I think
29:52
Speaker 3: this is true of a lot of different kinds of
29:53
Speaker 3: quantic computing modalities. Now the science has now been demonstrated
29:57
Speaker 3: error correction works. We need, like you said, a thousand
30:00
Speaker 3: times more cubits, and we want to go and do
30:03
Speaker 3: that by building network machines and scaling up the number
30:07
Speaker 3: of cubits inside each one of our systems and so on.
30:09
Speaker 3: And I'm sure our competitors have similar stories as well.
30:13
Speaker 5: Uh.
30:13
Speaker 3: And so the question just becomes, you know, is there
30:15
Speaker 3: the capital, is there the the excitement around building it?
30:19
Speaker 1: So you you mentioned your competitors. You are a private company, right.
30:26
Speaker 1: Your competitors include some public quantum computing companies, but they
30:30
Speaker 1: also include like Google and IBM, Right, like these giant
30:35
Speaker 1: companies with huge revenue streams and kind of all the.
30:40
Speaker 4: Money they want, literally but a lot of money. How
30:45
Speaker 4: do you compete? How do you how do how does that?
30:49
Speaker 4: How does that work for you? Oh?
30:52
Speaker 3: I mean I think that, uh, I mean like any industry.
30:55
Speaker 3: I mean I think there's you know, venture investors and
30:58
Speaker 3: stuff who are excited about, you know, first of all,
31:01
Speaker 3: building funding technology that's different than the technology that's found
31:05
Speaker 3: at those giant hyperscaler you know, big companies, and second,
31:09
Speaker 3: uh who want to invest in that technology because if
31:13
Speaker 3: that technology wins, they actually get significantly more back from
31:17
Speaker 3: their investment than if they invest in Google. Like if
31:20
Speaker 3: Google has a quantum computer. I actually don't think the
31:23
Speaker 3: Google quantum stock is going to double overnight. Whereas if
31:26
Speaker 3: all of a sudden we have a universal fault talent
31:28
Speaker 3: quantum computer at Adam, yeah, our value could go up
31:32
Speaker 3: by a ten or one hundred x or something like that.
31:34
Speaker 3: So I think there's just like an asymmetry there and
31:36
Speaker 3: returns things like that. Yeah.
31:39
Speaker 1: So one of the interesting things about the field is
31:43
Speaker 1: there are a bunch of different companies using very different approaches,
31:48
Speaker 1: like physically different approaches. And I recognize that maybe more
31:53
Speaker 1: than one will work, or some will be good for
31:54
Speaker 1: some things and something good for others. But there might
31:56
Speaker 1: be something of a binary outcome, right, Like somebody might win.
32:00
Speaker 1: Somebody might get there first. What do you think of
32:03
Speaker 1: the chances it'll be somebody other than you.
32:08
Speaker 3: I certainly think there's a chance. I think that who
32:13
Speaker 3: wins in the end is going to just come down
32:15
Speaker 3: to dollars per unit compute. Like I think there's someone
32:20
Speaker 3: will get there first. I hope it's Adam, but you
32:23
Speaker 3: know someone's going to get there first. I think there
32:25
Speaker 3: will be a few years of multiple people then getting there,
32:29
Speaker 3: multiple modalities getting there, And I think at the end
32:31
Speaker 3: of the day, what wins is just dollar per unit compute,
32:34
Speaker 3: And all of a sudden, if there's a ten x
32:36
Speaker 3: difference in price between running on this system versus running
32:40
Speaker 3: on another system, people are just going to gravitate towards
32:43
Speaker 3: a cheaper one. And I think that will be the
32:45
Speaker 3: one that actually turns out to be the one that
32:48
Speaker 3: you know, an entire industry is built around.
32:51
Speaker 1: Where do you think we should end the main part
32:52
Speaker 1: of this conversation, Like if you sort of sit back
32:55
Speaker 1: and think big thoughts and gaze off into the distance,
32:58
Speaker 1: like what like where do you land thinking about this today?
33:04
Speaker 3: I mean, I think the quantum computers are more of
33:06
Speaker 3: an inevitability than they've ever been. Like, I think that
33:09
Speaker 3: we've made such amazing progress as an industry that everyone
33:13
Speaker 3: is on board with the idea that no, no, this
33:15
Speaker 3: is this is just going to happen. Like if someone's
33:18
Speaker 3: going to have a universal fault talent quantum computer, people
33:21
Speaker 3: are going to be able to use it over the cloud.
33:23
Speaker 3: They're going to be able to do all the problems
33:25
Speaker 3: they want to do, and so on and so forth.
33:27
Speaker 3: And I feel like even I don't know, three four
33:29
Speaker 3: or five years ago, I don't think people would have
33:31
Speaker 3: said that across the board.
33:36
Speaker 1: We'll be back in a minute with the lightning round. Okay,
33:49
Speaker 1: let's finish with the lightning round. It's gonna be a
33:52
Speaker 1: little more random, but fun. I hope.
33:57
Speaker 4: Albert Einstein overrated or underrated?
34:00
Speaker 3: Underrated?
34:01
Speaker 1: Yeah, very highly rated. Richard Feynman overrated rounded, Uh.
34:13
Speaker 3: I think underrated. I think that actually seeing him give lectures.
34:16
Speaker 3: I didn't get to see him in person, but I mean,
34:19
Speaker 3: I think seeing recordings of him given lectures, I mean
34:21
Speaker 3: I think he was probably the most amazing physics teacher
34:24
Speaker 3: there was.
34:25
Speaker 4: Is it right? Did Feineman come up with the idea
34:27
Speaker 4: of the quantum computer?
34:28
Speaker 3: I read it also, Yeah, and he has this amazing
34:31
Speaker 3: sentence about how you know it would be so simple
34:33
Speaker 3: to do this with a bunch of atoms that you
34:36
Speaker 3: kind of rearrange and you kind of move around and
34:38
Speaker 3: stuff like that. Someone showed that to me many many
34:41
Speaker 3: years after I had started a company to do this
34:43
Speaker 3: with Adam.
34:43
Speaker 4: So was he right that it would be so simple?
34:46
Speaker 3: Absolutely? Yeah?
34:48
Speaker 4: Wait? What no? Is it right that.
34:53
Speaker 1: You made or helped to make an atomic clock that's
34:56
Speaker 1: precise to the second over five billion years?
34:59
Speaker 4: The most something precise clock ever, Yeah, so okay, A
35:04
Speaker 4: few questions following on that. Precisely? How early do you
35:08
Speaker 4: get to the airport to catch a plane? Oh?
35:11
Speaker 3: Actually, this is a great thing that someone taught me.
35:14
Speaker 3: If you don't miss a flight every like, you know,
35:17
Speaker 3: fifty times, you're always getting to the airport tour.
35:19
Speaker 1: Yes, that is a theory, that's an optimization. So what
35:22
Speaker 1: for you is the optimal fraction of flights to miss?
35:27
Speaker 3: I think like one percent or something like that.
35:29
Speaker 4: When was the last time you missed a plane?
35:32
Speaker 3: Probably like thirty forty fifty flights ago or something.
35:35
Speaker 4: So you're doing okay?
35:36
Speaker 3: Yeah ye.
35:45
Speaker 1: Ben Bloom is the co founder and CEO of Adam Computing.
35:50
Speaker 1: Today's show was produced by Gabriel Hunter Chang. It was
35:53
Speaker 1: edited by Lyddy jeene Kott and engineered by Sarah Bruguer.
35:57
Speaker 1: You can email us at problem at Pushkin dot FM.
36:01
Speaker 1: I'm Jacob Goldstein and we'll be back next week with
36:04
Speaker 1: another episode of What's Your Problem. I'm Jacob Goldstein, and
36:16
Speaker 1: right now we're going to play you a clip of
36:19
Speaker 1: a new show that I co host. The show's called
36:21
Speaker 1: Business History. My co host is Robert Smith and This
36:24
Speaker 1: clip is from an episode we did about how Nolan Bushnell,
36:29
Speaker 1: a stoner turned entrepreneur, created the video game company Atari
36:33
Speaker 1: and hired a young, inexperienced Steve Jobs. I really hope
36:37
Speaker 1: you like the clip, and if you want to hear more,
36:40
Speaker 1: you can find business history wherever you're listening to this
36:43
Speaker 1: show right now. This like nineteen year old hippie kid
36:47
Speaker 1: walks in and he says he won't leave until they
36:50
Speaker 1: give him a job, And the receptionist calls the head
36:53
Speaker 1: engineer and she goes, yeah, we got a hippie kid
36:57
Speaker 1: in the lobby.
36:57
Speaker 4: Says he won't leave until we hire him. Should we
36:59
Speaker 4: call the cops or let him in? And the engineer says,
37:03
Speaker 4: bring him on in.
37:04
Speaker 5: It's nineteen seventies, a Silicon valley, and this guy wanders
37:07
Speaker 5: in and he is when if you tell a story
37:10
Speaker 5: like this, you know who it is.
37:11
Speaker 4: It's Steve Jobs. It's Steve Job. So fun, it's so
37:14
Speaker 4: delightful and perfectly.
37:17
Speaker 1: Steve Jobs is very good at his job, yes, and
37:20
Speaker 1: very unpleasant to wrqu On.
37:23
Speaker 4: Surprising.
37:24
Speaker 1: He tells Bushnell that like everybody's soldering wrong, right, they're
37:27
Speaker 1: actually putting together the hardware sladering the hardware. He's probably right,
37:30
Speaker 1: and bush Nell's like, yeah, he was right. He keeps
37:32
Speaker 1: calling his manager a dumb shit.
37:34
Speaker 4: Probably was not.
37:35
Speaker 1: Kind, not necessary. Bushell winds up putting Jobs on the
37:40
Speaker 1: night shift, partly so he won't bother so many people,
37:43
Speaker 1: and partly because he knew that Jobs like to hang
37:46
Speaker 1: out at night with his buddy Steve Wozniak, who was
37:50
Speaker 1: a great engineer, would.
37:51
Speaker 4: Be great to have hanging around Atari.
37:53
Speaker 1: And in fact, Jobs and Wozniak helped to make Breakout
37:57
Speaker 1: a great a target.
37:58
Speaker 4: Remember breaking of.
37:59
Speaker 5: Course, you're trying to knock down the bricks in a wall.
38:01
Speaker 4: Still, games like that, when you got a little paddle
38:03
Speaker 4: at the bottom.
38:04
Speaker 5: End, there's a moment when it goes through the wall
38:05
Speaker 5: and then.
38:05
Speaker 4: Goes so good.
38:09
Speaker 1: So Jobs worked at Atari for a little while and
38:12
Speaker 1: then decided that he wanted to go off to India
38:14
Speaker 1: to find his guru. Perfect asked Atari to pay for
38:18
Speaker 1: the trip. Nobody ever said he lacked moxie, and wound
38:21
Speaker 1: up making a deal with Atari where they pay him
38:24
Speaker 1: to go part of the way there. They had exported
38:27
Speaker 1: some games to Germany and there was some kind of
38:28
Speaker 1: problem with the games in Germany and they're like, we'll
38:30
Speaker 1: send you to Germany to fix the games and then
38:33
Speaker 1: you can get the rest of the way to India.
38:35
Speaker 1: And you know, the Germans said Jobs was terrible to
38:37
Speaker 1: work with, but he fixed the games. I was thinking
38:41
Speaker 1: about like the link between Atarian Jobs and what did
38:43
Speaker 1: he learn there, And it felt like maybe a little overdetermined,
38:45
Speaker 1: but I do think you know, clearly he had this
38:49
Speaker 1: profound sense of aesthetics and of delight, right, Like think
38:53
Speaker 1: of the Macintosh, right, this breakthrough Apple machine in the eighties.
38:56
Speaker 1: It was round and instead of squares, it was rounded,
38:59
Speaker 1: and it cost them more money, but it was beautiful
39:01
Speaker 1: and it was fun and you were engaged with it
39:04
Speaker 1: like a game.
39:05
Speaker 5: It almost looked a little bit like an arcade game
39:07
Speaker 5: with the curves than that. I think that especially at
39:10
Speaker 5: Silicon Valley at the time when they were dealing in
39:13
Speaker 5: actual silicon Right, if you're making chips for somebody, you're
39:16
Speaker 5: thinking about the future and the computers. You're not thinking
39:19
Speaker 5: about the psychology of the customer, because the customer was
39:21
Speaker 5: another electronics company, right, and so Atari and the strength
39:25
Speaker 5: of it was really the first time where they're just like,
39:28
Speaker 5: how will a regular human being who has no training
39:31
Speaker 5: whatsoever interact with technology.
39:33
Speaker 1: Yeah, so Atari now mid seventies. They're selling all the
39:39
Speaker 1: machines they can make. They need more space, and so
39:43
Speaker 1: they rent an abandoned roller rink and they turn it
39:46
Speaker 1: into this office slash video game factory.
39:49
Speaker 5: I've been ten years older. I would have loved to
39:51
Speaker 5: have worked in a roller rink video game factory.
39:53
Speaker 4: Yes, everybody smoked weed. Yeah, there was a hot tub.
39:58
Speaker 1: There was a pool party where everybody ended up naked
40:00
Speaker 1: in the pool at Bushnell himself looked back on it
40:04
Speaker 1: later and said, if that isn't a horror show for
40:08
Speaker 1: any hr person today, I don't know what is