1
00:00:04,260 --> 00:00:11,740
The proposition is, it has to do with the acronym you heard several times today, NISQ.

2
00:00:12,840 --> 00:00:15,560
Noisy Intermediate Scale Quantum.

3
00:00:15,560 --> 00:00:19,400
You know we are in the NISQ era right now.

4
00:00:19,400 --> 00:00:33,640
And so the proposition for the village is, this village believes that the current state of NISQ is already enabling quantum advantage for those who know how to use these technologies.

5
00:00:34,240 --> 00:00:39,460
Arguing for, I'm sorry, yes, for the proposition is Rafal Janik.

6
00:00:39,460 --> 00:00:41,640
Other way around, sorry.

7
00:00:41,800 --> 00:00:44,540
Konstantinos, you're arguing for the proposition.

8
00:00:45,380 --> 00:00:48,200
Raf, you're arguing against the proposition.

9
00:00:48,520 --> 00:00:54,900
And I wanted to say this is so very important because we heard from both of them earlier today.

10
00:00:54,900 --> 00:00:57,420
You both know they know their stuff.

11
00:00:57,420 --> 00:01:00,200
And it is going to help us come to our conclusion.

12
00:01:00,300 --> 00:01:04,080
So first for the proposition, Konstantinos.

13
00:01:04,080 --> 00:01:05,720
If you like we can switch halfway.

14
00:01:05,720 --> 00:01:06,660
Yeah sure.

15
00:01:09,280 --> 00:01:20,980
So do we need fault-tolerant quantum computing to do anything useful?

16
00:01:22,100 --> 00:01:24,100
I don't think so.

17
00:01:24,280 --> 00:01:31,200
If there's one thing I've learned in this decade of quantum that I've been messing around, we get surprised all the time.

18
00:01:31,200 --> 00:01:34,860
So just real high level, first of all, we constantly are getting surprised.

19
00:01:34,860 --> 00:01:45,780
We find little performance edges here and there, little clever ways to do things like hybrid, where some of it's classical, some of it's quantum, and you put the two together and you see some kind of like performance edge.

20
00:01:46,280 --> 00:01:59,440
So I'm not gonna worry too much about like provable, benchmarkable advantage so much as like can you identify some step in a process that could be improved with quantum over classical.

21
00:01:59,440 --> 00:02:00,860
So I think we can.

22
00:02:00,860 --> 00:02:02,740
And I've already seen signs of it.

23
00:02:03,900 --> 00:02:07,380
Recently IBM did their error mitigation.

24
00:02:07,500 --> 00:02:18,120
So they were able to show that basically we were able to cancel out some noise by doing lots of runs and statistically removing it kind of accounted for it, just to keep it real simple.

25
00:02:18,360 --> 00:02:31,420
So it's error mitigation, but there's also error suppression, where you can understand how a particular machine works, and then, for example, you can go through and flip something to reverse an error before it starts to form.

26
00:02:33,640 --> 00:02:39,780
So these kinds of things could already show some benefit, and they did in the case of the IBM papers.

27
00:02:41,260 --> 00:02:48,540
So when you look at all those kinds of things, you're going to start to see some provable, I guess what I like to call a customer advantage.

28
00:02:48,540 --> 00:02:52,420
It's like when you look at everything and you apply it, you kind of get some kind of edge.

29
00:02:52,620 --> 00:03:04,220
I wrote a paper on doing portfolio optimization with cardinality constraints, and we were able to show some performance gain using D-Wave's hybrid solver and some approaches .

30
00:03:04,380 --> 00:03:06,700
So yeah, it's not a benchmarkable advantage.

31
00:03:06,700 --> 00:03:10,680
I'm not claiming that from rooftops, but we were able to show some gain to that customer.

32
00:03:10,980 --> 00:03:14,540
So I do think when you consider all those things, we're getting close.

33
00:03:14,540 --> 00:03:24,680
And one last thing I'll add, Quantinium recently, they did that whole thing with three, we'll call them logical in quotes, qubits that were able to simulate the energy states of an H2 molecule.

34
00:03:24,680 --> 00:03:26,140
So we're getting there.

35
00:03:26,140 --> 00:03:27,420
We're like, we're approaching.

36
00:03:27,420 --> 00:03:28,240
We're not going to go.

37
00:03:28,240 --> 00:03:32,100
So I think these little tweaks along the way will enable that.

38
00:03:32,220 --> 00:03:32,820
Awesome.

39
00:03:32,820 --> 00:03:33,820
Your rebuttal.

40
00:03:34,260 --> 00:03:35,860
Or your argument against it.

41
00:03:36,020 --> 00:03:40,860
I feel a little bit underprepared without memes on the screen behind me right now.

42
00:03:42,580 --> 00:03:47,940
So quantum computing has a lot of promise, and it's actually a really hard path to get there.

43
00:03:48,420 --> 00:04:01,900
Not only do I not believe that we're at the position right now where we're delivering on that promise, I'd also like to add that it's actually a distraction from the hard work that we have to do to focus on the NIS era.

44
00:04:02,080 --> 00:04:03,900
And there are a couple definitions to cover.

45
00:04:03,900 --> 00:04:05,760
The first one is quantum advantage.

46
00:04:05,920 --> 00:04:11,020
This is not being able to do something better than a classical computer.

47
00:04:11,020 --> 00:04:16,780
This is being able to do a meaningful task better than a classical computer.

48
00:04:17,200 --> 00:04:30,100
If anyone's curious, today while sitting over there preparing for this debate, I did a quick search and I found 41 instances of people claiming quantum advantage over the last 20 years.

49
00:04:30,140 --> 00:04:32,440
All of those have not come true.

50
00:04:32,440 --> 00:04:35,860
They really fall down for two reasons.

51
00:04:35,860 --> 00:04:39,960
One, they're flat-out fraudulent, which is probably the bulk of them.

52
00:04:39,960 --> 00:04:51,200
The other one comes down to the fact that classical algorithms are incredibly good and many quantum researchers are not classical algorithm researchers.

53
00:04:51,260 --> 00:05:02,920
So what you actually see, and historically we've seen this with multiple companies, is a claim of quantum advantage and two days later a proof that there's a classical method that is far better.

54
00:05:03,100 --> 00:05:06,460
Also, there's a lot of value in quantum computing today.

55
00:05:06,460 --> 00:05:08,180
I don't want to discount that.

56
00:05:08,240 --> 00:05:13,540
There's the value of what we're doing scientifically and all the research it's been able to unlock.

57
00:05:13,540 --> 00:05:20,400
There's the value around quantum adjacent technologies such as, you know, we've heard quantum inspired algorithms.

58
00:05:20,400 --> 00:05:25,460
All these are delivering value, but the one thing we have not seen is quantum advantage.

59
00:05:25,560 --> 00:05:37,340
And you would not believe how many companies, how many academic institutions, how many researchers only care about running a NISQ right now because they're able to run on hardware.

60
00:05:37,360 --> 00:05:51,720
Instead of thinking, how do we prepare for the hardware that's going to be here five years, ten years from now, moving past the NISQ era, not necessarily into fault-tolerant, but maybe to not noisy intermediate-scale quantum computers.

61
00:05:51,760 --> 00:05:59,480
Quantum computers that have gate fidelities and qubit fidelities that actually allow us to go deep enough in algorithms to show meaningful value.

62
00:05:59,800 --> 00:06:01,920
Great, we have ten minutes for rebuttals.

63
00:06:01,920 --> 00:06:03,420
So, your first rebuttal.

64
00:06:03,900 --> 00:06:10,380
Yeah, well, I mean, I don't disagree with anything you said, technically, so I don't know if I'm going to set this off on you.

65
00:06:12,120 --> 00:06:13,380
It's all good.

66
00:06:14,240 --> 00:06:22,460
I just think that we're not, you know, I wasn't bringing up any examples, obviously, of the advantage claims that were useless.

67
00:06:22,460 --> 00:06:23,220
I hate that.

68
00:06:23,220 --> 00:06:26,360
It's like, look, we did this thing that you're never going to do, but we did it better.

69
00:06:26,360 --> 00:06:27,420
It's like, okay, cool.

70
00:06:27,920 --> 00:06:30,820
It's just like, I don't like when someone writes code and then hacks it.

71
00:06:30,820 --> 00:06:32,240
That's kind of useless, too.

72
00:06:33,220 --> 00:06:36,860
You don't want to, like, write your implementation and hack it.

73
00:06:37,800 --> 00:06:46,340
So, I agree in that space.

74
00:06:46,340 --> 00:06:49,100
And that's why I got big time into quantum-inspired.

75
00:06:49,100 --> 00:06:52,600
You know, that's why attention networks, we're hoping to see some gains there.

76
00:06:52,600 --> 00:06:56,680
But I keep seeing that hybrid thing, you know, and I can't let it go.

77
00:06:56,680 --> 00:07:05,740
I just got this feeling that we're going to be able to take advantage of some of those hybrid learnings and apply them in the work stream.

78
00:07:05,740 --> 00:07:08,560
Because, like, very few problems are solved in one step, one thing.

79
00:07:08,800 --> 00:07:10,760
You know, like supply chain issues.

80
00:07:10,760 --> 00:07:13,660
There's, like, a million places where you could try and improve something.

81
00:07:14,540 --> 00:07:21,640
So, I just have a feeling in my gut that we're going to see the ability to tweak just a few of those the right way.

82
00:07:21,640 --> 00:07:24,980
And they're like, we're sort of, like, knocking on the door of that.

83
00:07:24,980 --> 00:07:27,760
And then there's also the machine technologies that are evolving so fast.

84
00:07:27,760 --> 00:07:29,800
So, I just go into it for a second.

85
00:07:30,180 --> 00:07:31,920
Like, neutral atom, for example.

86
00:07:32,180 --> 00:07:36,980
We're talking high qubit counts, really long lifetimes.

87
00:07:37,000 --> 00:07:47,480
Like, we're expecting the next generation of neutral atom machines that are coming out within a year or so to have lifetimes of, like, I don't know, tangling times of 10 seconds or so.

88
00:07:47,480 --> 00:07:53,340
10 seconds is like, you know, the age of the universe, practically, when you're thinking about a quantum computer, you know.

89
00:07:53,580 --> 00:07:57,640
So, if we have a tangle time of 10 seconds, it makes things a little more interesting.

90
00:07:58,320 --> 00:08:01,640
Right now, they're running what's called analog mode, which is, like, not quite.

91
00:08:01,640 --> 00:08:04,540
You can't run gates on them, so it's a little closer to annealing than it is to gates.

92
00:08:04,540 --> 00:08:05,880
But they will be running gates.

93
00:08:05,940 --> 00:08:12,680
And they're gonna have modes where you're gonna be able to sacrifice a few extra qubits to get, like, better fidelity when needed.

94
00:08:12,840 --> 00:08:25,560
So, whenever you think about those, like, out-of-the-box approaches, I just have a feeling we're gonna see, like, at least a few of the problems show, like, some kind of advantage relatively soon, without true, you know, fault tolerance.

95
00:08:26,340 --> 00:08:26,920
Great.

96
00:08:26,920 --> 00:08:28,120
Rob, your rebuttal.

97
00:08:28,740 --> 00:08:34,640
I mean, this might get into a weird debate because we're gonna start agreeing with each other more than disagreeing.

98
00:08:34,980 --> 00:08:38,600
But I'd like to point out a few things that were brought up a moment ago.

99
00:08:39,080 --> 00:08:41,340
Longer lifetimes, better fidelities.

100
00:08:41,340 --> 00:08:43,900
This actually pushes us out of the misc era.

101
00:08:44,240 --> 00:08:52,940
Everything that we're talking about doing, that we want to do, requires much longer lifetimes, much better fidelities.

102
00:08:52,940 --> 00:09:00,840
We know a handful of algorithms, and you can count them on one hand today, that can prove quantum advantage.

103
00:09:02,080 --> 00:09:03,760
That's a little bit worrying.

104
00:09:03,760 --> 00:09:16,720
Instead of worrying about which optimization problem can we try to jam onto a processor today, we should be worrying about how do we actually expand the number of algos that can provide quantum advantage in the future.

105
00:09:17,440 --> 00:09:24,820
Maybe even a little more scary than that is the fact that only one of those provides exponential advantage.

106
00:09:25,260 --> 00:09:32,980
And there's been papers out of Google AI in the past that demonstrate, even with fault tolerance, quadratic advantage might not be enough.

107
00:09:33,000 --> 00:09:46,000
The overheads to build a quantum computer, the way that you actually perform the error correction, might require much higher speedups than quadratic in order to actually overcome the overhead.

108
00:09:46,060 --> 00:10:00,000
So I would encourage everyone again to not worry about jamming your algo today, your H2 molecule, which you can simulate on your laptop and calculate the ground state on, without needing a quantum computer.

109
00:10:00,000 --> 00:10:15,080
But actually start focusing on understanding how do you improve the readout method for a quantum computer beyond the quantum phase estimation, which pretty much every quantum chemistry algo is based on today.

110
00:10:15,340 --> 00:10:31,020
We have a few more minutes for rebuttals, but those of you just coming in, I want you to know what we're debating is the state of current quantum computers and if the state of current quantum computers is good enough to give advantage, real strategic advantage,

111
00:10:31,020 --> 00:10:32,920
to those who really know the technology.

112
00:10:32,920 --> 00:10:36,280
We're arguing for and against and then we're in the middle of our rebuttals.

113
00:10:36,280 --> 00:10:39,400
In a few minutes we'll get to audience questions and answers and inputs.

114
00:10:39,500 --> 00:10:43,340
So we have two more minutes for final comments before we get to questions and answers.

115
00:10:43,360 --> 00:10:45,860
In keeping with the theme, I agree with everything you said.

116
00:10:47,960 --> 00:10:48,640
Guys!

117
00:10:49,060 --> 00:10:50,140
I'm sorry.

118
00:10:51,860 --> 00:10:53,400
We agree but disagree.

119
00:10:53,400 --> 00:10:53,800
Yeah .

120
00:10:54,400 --> 00:11:01,980
But like you said about not worrying about doing a thing that you can simulate on your laptop or whatever, the whole point is like the way it was done, right?

121
00:11:01,980 --> 00:11:08,100
You know, just like the first time we ran Schumer's algorithm, it wasn't that we couldn't do 15 is 5 times 3.

122
00:11:08,100 --> 00:11:09,300
I mean, you could do that.

123
00:11:09,300 --> 00:11:10,780
It's just the way it was run.

124
00:11:10,780 --> 00:11:21,340
So like the fact that we're approaching fault tolerance through these logical qubits, that we're doing things like error suppression, mitigation, like these are like the building blocks.

125
00:11:21,340 --> 00:11:23,940
So I just view them as incredibly important for that purpose.

126
00:11:23,940 --> 00:11:31,800
Even if we could still run the math on a classical machine, I feel like without those building blocks being proven, we're not going to get far enough fast enough.

127
00:11:31,800 --> 00:11:35,500
I actually like the Shor example quite a bit.

128
00:11:36,080 --> 00:11:42,500
So I think I had the meme up earlier, which was factoring 15 into its primes versus the promise of Shor's.

129
00:11:42,500 --> 00:11:51,740
When Shor's first came out, the implementation actually would never run on a physical quantum computer with any imperfections.

130
00:11:51,740 --> 00:11:53,620
You needed the idealistic machine.

131
00:11:53,920 --> 00:12:11,260
And what I would say has been much more important to the development of actually running Shor's has been the work done from the top down, bringing the overheads required for Shor's down, instead of worrying about can you demonstrate factoring 15 on a quantum computer.

132
00:12:11,260 --> 00:12:18,220
It's again a great flex, I would say, to do that and to show that, but it's not doing actual meaningful work.

133
00:12:18,500 --> 00:12:37,760
One final point that can probably prove that there is no real quantum advantage, and it's the fact that the people like myself and the folks from Oxford Quantum Circuits are not locked away in a basement right now, working for whichever government that they happen to be under the control of.

134
00:12:37,760 --> 00:12:46,780
Because that really will happen as we get closer and closer to true quantum advantage, because the regulations just haven't caught up to where the technology is headed.

135
00:12:48,480 --> 00:12:50,720
Yeah, I was just thinking they're going to steal the machines.

136
00:12:50,720 --> 00:12:52,280
I didn't want them to steal you guys too.

137
00:12:52,280 --> 00:12:54,260
I think we go with them.

138
00:12:54,260 --> 00:12:56,120
There's no separating the two.

139
00:12:56,200 --> 00:13:00,180
And to be fair, it won't be me, but those two guys are definitely in trouble.

140
00:13:01,020 --> 00:13:05,920
I know there's no true quantum advantage right now, because everyone would know about it.

141
00:13:05,920 --> 00:13:07,440
It would be big news.

142
00:13:07,440 --> 00:13:13,060
But I do believe that we're going to have actual usable use cases.

143
00:13:13,060 --> 00:13:19,360
I just think we're going to have them before we're proclaiming true fault-tolerant intense gate caps.

144
00:13:20,560 --> 00:13:22,940
And then watch out for those oddball developments.

145
00:13:23,720 --> 00:13:25,700
Every once in a while, one of them is going to come true.

146
00:13:25,700 --> 00:13:30,980
Maybe not that Chinese paper that you need 327 qubits to crack RSA.

147
00:13:30,980 --> 00:13:32,240
Maybe not that one.

148
00:13:32,400 --> 00:13:34,440
But something like that might happen too.

149
00:13:34,860 --> 00:13:36,940
That's the one thing I just want to remind you of.

150
00:13:36,940 --> 00:13:39,840
These surprises are just things we never would have thought of.

151
00:13:40,240 --> 00:13:41,320
I keep seeing things.

152
00:13:41,320 --> 00:13:42,680
I'm like, I didn't see that coming.

153
00:13:42,680 --> 00:13:43,220
Cool.

154
00:13:43,460 --> 00:13:45,960
So that's one of the greatest things about working quantum.

155
00:13:47,760 --> 00:13:48,340
Thanks.

156
00:13:48,340 --> 00:13:51,360
And now it's time to transition to your inputs before we vote.

157
00:13:51,360 --> 00:13:57,960
You get an opportunity to ask questions, make comments, push back, make your own arguments, and then we will vote on the proposition.

158
00:13:57,960 --> 00:13:58,740
I saw yours first.

159
00:13:58,740 --> 00:13:59,140
Please, sir.

160
00:13:59,700 --> 00:14:05,000
So what do you think is holding up the development of those quantum algorithms?

161
00:14:05,000 --> 00:14:09,820
Is it that we don't have the right people trying to build the quantum algorithms right now?

162
00:14:09,820 --> 00:14:15,600
Or that we're lacking hardware to play around with to validate those, the simulation capabilities?

163
00:14:16,100 --> 00:14:16,960
What are your thoughts?

164
00:14:16,960 --> 00:14:21,760
All right, just to repeat the question, it is, what do you think is holding up the development of these quantum capabilities?

165
00:14:22,380 --> 00:14:23,400
Both of you, please.

166
00:14:23,400 --> 00:14:28,380
I think broadly, I would say tooling.

167
00:14:28,600 --> 00:14:32,720
And I know it might be a very simple answer, but it actually encompasses quite a bit.

168
00:14:33,360 --> 00:14:37,600
You know, somebody mentioned it's like being in the 1940s right now, quantum technology.

169
00:14:37,680 --> 00:14:40,680
I'd say that that's generous where we are.

170
00:14:41,020 --> 00:14:44,780
We're hoping we're going to move more quickly compared to digital computing.

171
00:14:44,900 --> 00:14:51,940
But it really is like, you know, before the development of digital computing, trying to come up with potential algorithms.

172
00:14:53,300 --> 00:14:57,540
We don't have the software tooling to allow researchers to do what they need to do.

173
00:14:57,540 --> 00:15:00,840
We don't have the hardware tooling for them to be able to test it out.

174
00:15:01,680 --> 00:15:08,380
And unfortunately, we also don't have the person power in order to actually go after this very aggressively.

175
00:15:09,560 --> 00:15:15,360
You know, I'll plug a little bit of the work that the team at the company I work at did, Xanadu.

176
00:15:16,360 --> 00:15:27,860
Looking at the simulation of batteries, again, not on today's computers, but in ones that might exist in a decade or in five years from now, they reduced the gate count by a factor of 40,000.

177
00:15:28,300 --> 00:15:28,840
Right?

178
00:15:28,960 --> 00:15:31,140
And this is, I would say, low-hanging fruit.

179
00:15:31,140 --> 00:15:31,940
They might disagree.

180
00:15:31,940 --> 00:15:33,280
That was a lot of hard work.

181
00:15:34,300 --> 00:15:36,480
But these are the type of gains that we can make.

182
00:15:36,480 --> 00:15:40,560
Unfortunately, the total gate count that you need is still a ways away.

183
00:15:40,660 --> 00:15:49,520
But we should be pushing that from the top down as we push the hardware from the bottom up until they finally converge in some place where we can deliver something meaningful.

184
00:15:51,120 --> 00:15:55,000
Yeah, some of this is theoretical, right?

185
00:15:55,000 --> 00:15:59,880
If you think about it, the greatest algorithms of all time were developed without quantum gear, right?

186
00:15:59,880 --> 00:16:00,700
They just work.

187
00:16:01,020 --> 00:16:03,040
You know, you go right back to the very first one.

188
00:16:03,500 --> 00:16:05,600
Well, they're still the greatest of all time.

189
00:16:06,500 --> 00:16:09,860
So they were in 1994, 1996, we had nothing.

190
00:16:09,860 --> 00:16:11,140
You know, they were just a dream.

191
00:16:12,160 --> 00:16:16,540
So I do think at some level it's going to be ingenuity, you know?

192
00:16:16,600 --> 00:16:19,280
And it might not even be human ingenuity, right?

193
00:16:19,280 --> 00:16:24,700
Like, for all we know, we'll get sparks of AGI in a quantum vein.

194
00:16:24,700 --> 00:16:31,860
You know, you'll be like, hey, insert name of LLME love, write me a new quantum algorithm, and it might be possible.

195
00:16:32,880 --> 00:16:33,920
So there's that too.

196
00:16:33,920 --> 00:16:35,120
It might not be us.

197
00:16:35,360 --> 00:16:40,640
Or AGI might solve some of the engineering hurdles and things.

198
00:16:40,640 --> 00:16:49,740
Not to turn this into a whole LLM discussion, but just because of what I've seen with, like, working with tensor networks and compressing them, I think there's a lot more potential there that could be tapped.

199
00:16:49,740 --> 00:16:51,140
So I would like to see that.

200
00:16:51,140 --> 00:16:55,300
So I think what we're missing is also that we're just limiting it to human beings.

201
00:16:55,360 --> 00:16:57,360
And it might be, you know, one of our friends.

202
00:16:57,360 --> 00:16:59,140
Before they kill us, they might help us.

203
00:17:01,460 --> 00:17:02,060
Great.

204
00:17:02,060 --> 00:17:03,000
There's a comment here.

205
00:17:03,000 --> 00:17:03,660
Thanks.

206
00:17:04,440 --> 00:17:11,280
So in your definition of quantum advantage, you mentioned meaningful tasks.

207
00:17:12,360 --> 00:17:19,620
And you also mentioned, I heard, go around real strategic advantage and something meaningful.

208
00:17:19,680 --> 00:17:31,460
I want to know what were the benchmarks and the measurements that determine what a meaningful task is, rather than the terms kind of sounding like they are shrouded by subjects.

209
00:17:31,940 --> 00:17:35,520
Good question about what is a meaningful task in your mind.

210
00:17:36,080 --> 00:17:38,940
I can tell you what not a meaningful task is.

211
00:17:38,940 --> 00:17:50,420
And it's an achievement I'm very proud of that, again, Google demonstrated first, but we were able to achieve at Xanadu, which is something called quantum supremacy or quantum computational advantage.

212
00:17:50,420 --> 00:18:09,120
And this is cherry picking a mathematical task that a quantum computer is really good at, but classical computers are very bad at, but that you can demonstrate mathematically from first principles that you've done with a quantum computer and it would take you millions of years,

213
00:18:09,120 --> 00:18:14,920
hundreds of years, whatever it is, but much longer with all the computational power in the world to deliver.

214
00:18:14,920 --> 00:18:18,840
And this has been demonstrated now five times, I believe, maybe four times.

215
00:18:19,360 --> 00:18:23,020
Right, on an optical and a photonic and a supercomputer.

216
00:18:23,120 --> 00:18:30,160
Great engineering milestones, whether they're necessary to get to where you're going, it depends on your your roadmap.

217
00:18:30,420 --> 00:18:35,040
Now, quantum advantage, it's not a very well defined term.

218
00:18:35,040 --> 00:18:39,320
In fact, none of these terms are, they kind of exist out in the zeitgeist.

219
00:18:39,760 --> 00:18:43,040
But what I would say is, it comes down to money.

220
00:18:43,040 --> 00:18:51,680
If it's something that somebody is willing to pay you for, and somebody doesn't show up the next day and say, I could do this faster classically.

221
00:18:51,680 --> 00:18:52,660
So let me rephrase that.

222
00:18:52,660 --> 00:18:58,820
If somebody who is well informed is willing to pay you for it, then it's real quantum advantage.

223
00:18:58,960 --> 00:19:06,700
And very honestly, it's not something we've seen yet from any quantum company, unless they've achieved it and have been incredibly secret.

224
00:19:08,020 --> 00:19:09,100
Yeah.

225
00:19:09,380 --> 00:19:15,180
So with quantum advantage, right, you're just, it has to be useful.

226
00:19:15,200 --> 00:19:17,400
It's moving the goal post argument.

227
00:19:17,400 --> 00:19:21,940
So some scientists don't like it because they're like, no, no, no, if quantum could do better than classical, that's advantage.

228
00:19:22,060 --> 00:19:24,340
But again, it's a useless problem.

229
00:19:24,340 --> 00:19:30,140
Are you just inventing some, you know, a hamster running in a wheel for no purpose, and quantum can run faster?

230
00:19:30,140 --> 00:19:31,520
It's kind of useless, right?

231
00:19:31,680 --> 00:19:34,840
So that's sort of what we've had so far, unfortunately.

232
00:19:35,300 --> 00:19:40,140
To get some of the advantage like I'm talking about, we've had to do other things too.

233
00:19:40,140 --> 00:19:47,060
Like I talked about tensor networks before, like, you know, we've used those to compress neural networks and then layer them in and et cetera.

234
00:19:47,060 --> 00:19:51,960
Like you get some other little bits of advantage along the way, and then you get your performance increases.

235
00:19:51,960 --> 00:19:54,080
But yeah, I can't claim we have quantum advantage.

236
00:19:54,360 --> 00:20:11,340
One thing to be careful of when you see these claims, often people will say this can't be simulated as a claim to prove the power of the quantum computer.

237
00:20:11,380 --> 00:20:14,980
But we can't simulate on a quantum level this table, right?

238
00:20:14,980 --> 00:20:19,440
But we don't have the control over it to actually make it do anything useful.

239
00:20:19,520 --> 00:20:24,240
So a lot of claims kind of fall short when it comes to the hardware.

240
00:20:24,240 --> 00:20:31,820
I think somebody mentioned a large-scale entangling coming out of China, a number of qubits that if they are fully entangled... 51.

241
00:20:31,820 --> 00:20:33,160
Yeah, it's very impressive.

242
00:20:33,160 --> 00:20:42,740
But now the next question that needs to be asked, can they exert full control over those 51 qubits to actually do something useful?

243
00:20:42,740 --> 00:20:46,980
Because if they can't, then I mean it comes down to the thing that I can point to.

244
00:20:46,980 --> 00:20:50,120
Look at my five million entangled qubits.

245
00:20:50,260 --> 00:20:54,580
Don't worry for how long they're entangled or whether I can control them or not, but they're right here.

246
00:20:54,880 --> 00:20:57,680
I'm pretty sure if you had five million entangled qubits, it doesn't matter a lot.

247
00:20:57,680 --> 00:21:00,040
I'm pretty sure it's right here.

248
00:21:00,620 --> 00:21:02,520
There's five million qubits right here.

249
00:21:04,900 --> 00:21:05,520
Questions?

250
00:21:05,520 --> 00:21:06,460
Yes, here, please.

251
00:21:06,480 --> 00:21:14,020
Basic question here, in part because you've already covered this, I came a little late, but what are the consequences of breaching quantum tech?

252
00:21:14,020 --> 00:21:24,940
I mean, you know, everybody talks about the advantages we have once we reach that stage, right, of using it, maybe commercializing it or something, but what are the consequences, the cold, like the cold hard truth of what it means?

253
00:21:26,880 --> 00:21:33,340
Also, to the gentleman on the right, did anybody tell you you look like Sheldon from the Big Bang Theory?

254
00:21:35,600 --> 00:21:37,040
It's way cooler.

255
00:21:37,460 --> 00:21:42,560
Yeah, I've been accused of talking like people don't understand what I'm saying.

256
00:21:43,240 --> 00:21:49,980
All right, so the question, to repeat the question, it's about what happens when we reach this state of a significant quantum advantage.

257
00:21:50,660 --> 00:21:53,680
Yeah, so a couple things will happen, right?

258
00:21:54,220 --> 00:22:05,560
If we have advantage and a large enough scale machine, we're at DEF CON, so there's the obvious thing that's going to happen, major cryptographic problems, and that all begins next year, by the way.

259
00:22:05,560 --> 00:22:11,660
NIST is going to publish this new standard, so starting next year, that's when the quantum apocalypse begins, because you have to start fixing it.

260
00:22:11,660 --> 00:22:16,220
Once the government says this is deprecated, you've got to fix it, whether you believe in quantum computers or not.

261
00:22:16,320 --> 00:22:18,320
So the apocalypse is here anyway.

262
00:22:18,320 --> 00:22:21,560
So that's going to happen one way or another, if you're the person fixing this stuff.

263
00:22:21,640 --> 00:22:25,900
What doesn't happen is we're going to start solving problems that we never thought we'd solve before.

264
00:22:26,380 --> 00:22:28,720
I think they're going to use less energy.

265
00:22:29,780 --> 00:22:39,560
Even though you imagine these refrigerators and all that, still, a quantum computer does a whole lot in a very quick period of time, with very little actual energy being expended.

266
00:22:39,720 --> 00:22:42,000
So that's the major thing that's going to happen.

267
00:22:42,000 --> 00:22:45,560
We're going to have these huge problems being solved that we were never able to tackle before.

268
00:22:45,560 --> 00:22:47,440
Things may be involved in the climate.

269
00:22:48,020 --> 00:22:50,400
Modeling chemicals that can take carbon out of the air.

270
00:22:50,400 --> 00:22:50,920
We don't know.

271
00:22:50,920 --> 00:22:54,360
It's like the use cases are almost unlimited when you get to that level.

272
00:22:54,360 --> 00:22:56,940
So it's a really exciting thing to look forward to.

273
00:22:57,640 --> 00:22:58,960
So is it a series?

274
00:23:00,300 --> 00:23:03,880
No, because I think we're going to start preparing for the cryptography.

275
00:23:03,880 --> 00:23:09,600
And Chrome is about to add, in version 116, they're about to add their hybrid post-quantum to that.

276
00:23:09,600 --> 00:23:12,280
They've already had it in AWS.

277
00:23:12,560 --> 00:23:14,780
So these things are happening already.

278
00:23:14,780 --> 00:23:17,920
I think if we get ahead of it, it'll be like with Y2K.

279
00:23:17,920 --> 00:23:18,980
Nothing happened.

280
00:23:19,120 --> 00:23:21,220
It was like, oh crap, it's okay.

281
00:23:21,260 --> 00:23:22,820
So it might be fine.

282
00:23:22,820 --> 00:23:23,760
Are you sure?

283
00:23:24,880 --> 00:23:26,240
That nothing happened?

284
00:23:29,040 --> 00:23:32,640
There are a couple applications that people think will come first.

285
00:23:33,300 --> 00:23:35,780
It was mentioned, quantum chemistry.

286
00:23:35,780 --> 00:23:47,640
So basically, one thing that a quantum computer is most likely good at, and again, we don't really know this until we're able to actually run it and prove it, but it's at simulating quantum systems.

287
00:23:47,980 --> 00:23:59,740
So this means things like, you know, building better batteries, because now you can simulate all the different electrolytes you would want to without actually have to go through and manufacture a battery.

288
00:24:00,140 --> 00:24:01,980
It was mentioned the climate impact.

289
00:24:01,980 --> 00:24:11,640
So this is usually around catalysts, usually based around transition metals that are very hard to simulate on a quantum level right now using traditional methods.

290
00:24:11,880 --> 00:24:21,500
But would require a modest, still a ridiculous compared to NISQ, but a modest for a quantum quantum computer number of cubits in order to actually go backward.

291
00:24:21,920 --> 00:24:23,300
But I'd say we don't know.

292
00:24:23,300 --> 00:24:29,060
It's like sitting in, you know, 1920 and saying, hey, what are we going to do with digital computers?

293
00:24:29,440 --> 00:24:34,720
I don't think we have any idea what, like, the equivalent of the quantum tender application is going to be.

294
00:24:34,860 --> 00:24:38,980
It's going to take, you know, a good amount of time to get there.

295
00:24:38,980 --> 00:24:40,660
We'll have a lot of thoughts, though.

296
00:24:42,240 --> 00:24:51,460
We have time for a few more questions before we vote on the proposition that they have been debating on this issue of quantum compute today.

297
00:24:51,460 --> 00:24:55,380
Is there any advantage by really knowing them?

298
00:24:55,500 --> 00:24:58,280
So here, please, first, and then we'll come back to you.

299
00:24:59,460 --> 00:25:18,080
What is your perspective on the investment of the relationship of quantum and quantum technology as a whole, including computing and energy as resources dwindle for building batteries and building computers in general?

300
00:25:18,080 --> 00:25:19,780
Do you see investment?

301
00:25:19,780 --> 00:25:22,620
How do you see that investment happening?

302
00:25:22,620 --> 00:25:28,160
Discuss, please, the investment and the opportunity between energy and quantum.

303
00:25:28,680 --> 00:25:29,520
Sure.

304
00:25:30,280 --> 00:25:39,940
So from the energy perspective, quantum computers don't operate faster than classical computers.

305
00:25:39,940 --> 00:25:42,120
They don't process everything in parallel.

306
00:25:42,120 --> 00:25:46,320
The best example is leveraging quantum mechanics.

307
00:25:46,320 --> 00:25:55,620
You can essentially encode your problem, your very hard problem, on a very complicated system and control it to get the answer.

308
00:25:55,620 --> 00:25:58,840
So you're sidestepping the difficult calculation.

309
00:25:58,840 --> 00:26:03,740
By difficult, I mean completely intractable by classical computation.

310
00:26:03,780 --> 00:26:07,380
And this is what all algorithms, quantum algorithms, are based on.

311
00:26:07,380 --> 00:26:08,780
So the hope is huge.

312
00:26:08,780 --> 00:26:15,760
The hope for, you know, resource scarcity is really massive.

313
00:26:15,760 --> 00:26:25,220
And it comes down to anything from optimization work, to quantum chemistry, to developing small molecule pharmaceuticals, and then all the way, all the way beyond that.

314
00:26:25,260 --> 00:26:31,840
Maybe if I understood your second part of the question was, like, what is the impact of manufacturing these things?

315
00:26:31,840 --> 00:26:34,600
Like, these things actually need to be made out of different materials.

316
00:26:34,860 --> 00:26:37,100
It really depends on the modality.

317
00:26:37,460 --> 00:26:39,360
Some things are really straightforward.

318
00:26:39,360 --> 00:26:41,720
They're silicon-based.

319
00:26:41,720 --> 00:26:47,060
There's, you know, no real super rare components to it.

320
00:26:47,060 --> 00:26:50,940
Other ones are based on much more scarce things.

321
00:26:51,140 --> 00:27:01,620
But I would say the impact that they will have is going to be so much greater than the resources that they're going to consume, compared to what classical computers are able to do today.

322
00:27:02,980 --> 00:27:05,980
And some of them could be made with, like, literally nothing.

323
00:27:05,980 --> 00:27:08,840
I mean, ions are, they're free.

324
00:27:08,840 --> 00:27:09,420
They're everywhere.

325
00:27:09,420 --> 00:27:10,200
You don't have to make them.

326
00:27:10,200 --> 00:27:11,100
They're already there.

327
00:27:11,520 --> 00:27:12,980
I don't know about ytterbium.

328
00:27:13,560 --> 00:27:15,060
Ytterbium, you know, whatever.

329
00:27:16,520 --> 00:27:17,300
There's some in here.

330
00:27:17,300 --> 00:27:18,580
I promise you there's some in here.

331
00:27:18,580 --> 00:27:20,020
I'm going to find it when we're done.

332
00:27:20,020 --> 00:27:20,840
I'll show you.

333
00:27:21,260 --> 00:27:22,340
It'll glow in my hands.

334
00:27:23,980 --> 00:27:28,960
So, yeah, so some of it depends, like you said, on the modality, what they're made of.

335
00:27:28,960 --> 00:27:36,100
And we're already using quantum computers in, like, simple optimization things for, like, how to wrap power, things like that.

336
00:27:36,100 --> 00:27:42,420
So these are, like, hybrid problems we're trying to solve, using some annealing and some other steps, and tensorizing, whatever.

337
00:27:42,440 --> 00:27:46,700
So we're very concerned with, like, using, applying this to power in general.

338
00:27:46,700 --> 00:27:48,000
Because, yeah, we've got a huge problem.

339
00:27:50,020 --> 00:27:51,800
One more question, then we vote.

340
00:27:52,620 --> 00:27:56,940
In terms of academia, let me just give you a 20 seconds context.

341
00:27:57,320 --> 00:28:04,120
ACM, IEEE, AAAI, Computer Science Curricula is the project that they're working for the next 10 years.

342
00:28:04,140 --> 00:28:04,780
Curricula.

343
00:28:04,780 --> 00:28:15,240
They're identified 18 knowledge areas, algorithms, operating systems, HCI, all the way to SEP, which is Society Ethics and Professionalism.

344
00:28:15,260 --> 00:28:16,780
You know, we're in DEF CON.

345
00:28:16,780 --> 00:28:17,740
We care about cyber.

346
00:28:17,740 --> 00:28:25,440
But at the same time, you're talking about green technologies and other aspects that quantum will have a big impact for the next 10 years.

347
00:28:25,440 --> 00:28:35,340
I was a little bit disappointed that the curriculum guidelines, they only have, like, less than two percent of learning outcomes and competencies in quantum.

348
00:28:35,340 --> 00:28:36,620
However, they're there.

349
00:28:36,760 --> 00:28:42,680
But, for example, cybersecurity is infused in the 18 knowledge areas.

350
00:28:42,680 --> 00:28:55,820
If I will ask you, and this is a question, if I ask you, what would be the top three computing areas that you should focus curriculum, like, heavily curriculum, what would those be in your experience?

351
00:28:55,820 --> 00:29:03,980
All right, what are the top three curriculum that you would focus on for computer science education overall?

352
00:29:06,160 --> 00:29:20,780
I would say, it's pretty straightforward for me, it would be building better quantum systems, writing better quantum algorithms, and absolutely machine learning, too, because it intersects so much.

353
00:29:20,780 --> 00:29:21,700
So, those are the three.

354
00:29:22,500 --> 00:29:24,720
I'm going to say the same way, but different.

355
00:29:24,720 --> 00:29:27,660
Linear algebra, linear algebra, and linear algebra.

356
00:29:30,460 --> 00:29:33,360
Yeah, of course.

357
00:29:33,360 --> 00:29:35,720
That's like saying you need the alphabet before you write something.

358
00:29:38,080 --> 00:29:41,440
I tried so hard to, like, keep linear algebra to a minimum.

359
00:29:45,060 --> 00:29:46,860
Ladies and gentlemen, it's time to vote.

360
00:29:46,860 --> 00:29:56,680
So, what I'm going to do is I'm going to say the proposition again, and then we will ask you to raise your hands if you support the proposition based on the arguments that you just heard.

361
00:29:56,900 --> 00:30:03,600
We are in the era of noisy intermediate scale intermediate scale quantum NISQ.

362
00:30:03,600 --> 00:30:05,260
This is the NISQ era.

363
00:30:05,440 --> 00:30:07,620
NISQ and the future of quantum advantage.

364
00:30:07,620 --> 00:30:16,840
This village believes the current state of NISQ is already enabling quantum advantage for those who know how to use these technologies.

365
00:30:16,840 --> 00:30:19,140
If you support that, raise your hand

366
00:30:22,230 --> 00:30:22,830
.

367
00:30:23,770 --> 00:30:25,570
I don't think it's already.

368
00:30:27,830 --> 00:30:29,470
That's the version I have.

369
00:30:30,570 --> 00:30:32,710
There's a lot of undecided.

370
00:30:32,870 --> 00:30:40,050
Yeah, if I could rephrase it, I would say that we don't need fault-tolerant for advantage, but I'm not saying we have an advantage now, for sure.

371
00:30:41,950 --> 00:30:43,790
Mark, how did that vote go, do you think?

372
00:30:44,130 --> 00:30:45,470
Oh, definitely against.

373
00:30:47,110 --> 00:30:50,590
Thank you for the three people who were reliably sort of going, yeah, sure.

374
00:30:52,730 --> 00:30:54,630
Well, we have captured this.

375
00:30:54,630 --> 00:30:59,650
I know there were a lot of other questions, so please, again, join our Discord server and keep asking your questions there.

376
00:30:59,650 --> 00:31:02,130
And with that, I want to say thanks to our debaters.

377
00:31:05,410 --> 00:31:06,670
Thanks, Mark.

378
00:31:06,810 --> 00:31:07,630
Thank you.