Helen Toner On Chinese Growth, AI Milkshakes And Making Bets


James: Hey, everyone. Welcome back to Browser Bets. Today I'm with Helen Toner, the Director of Strategy at Georgetown Center for Security and Emerging Technologies. She's also on the board of OpenAI and has a wealth of knowledge and experience both about China, technology and artificial intelligence and we're so excited to have her on the show with us today. Anything I missed in your bio there?

Helen: No, I think that covered the basics.

James: Awesome. So, you have a little bit of experience during forecasting as a part of your professional experience. What do you want to forecast with us today?

Helen: Yeah, I am a huge fan of forecasting and using bets as a way to make our opinions, beliefs and ideas about the world more concrete and make it easier to engage with them. I work in Washington so there's a lot of social pressure to be an expert on everything.

On Chinese economics and the relative power of the US

James: Fair enough. Let's dive into the first one and talk about Chinese economics. I'm a lay observer but what do you think people should know? What are the trends that are interesting to predict?

Helen: My day-to-day work is national security focused, so the big one that comes up here is just the relative power of the US and China, the rise of China, and China becoming economically and militarily much more powerful than it has been before.

There's some stuff we can already see there. If you're just looking at straight up GDP numbers, for instance, the US has its line that's been going up and China's line has been going up. And, not to get too nerdy, but if you look at one measure of GDP which is adjusted for purchasing power -- so saying, "Well, it turns out lots of stuff is cheaper in China so their GDP is effectively higher" -- if you adjust for purchasing power, China already passed the US in terms of GDP, around 2016, somewhere in that vicinity.

If you don't adjust your purchasing power and you just look at nominal GDP, which is literally the number of US dollars without any adjustments, then the US still has higher GDP and China's slowly closing in and I think is expected to crossover in the next 10-ish years. Again, this would be something I think various statistical agencies would have very high quality projections of.

I'm not sure what could be interesting to try and bet on there. I think literal GDP forecasts, I would just go look at the statistical agencies and see what they say and be like, "Yeah, probably that's right." Like the World Bank or, I'm not even sure who would actually have the right stats there, but experts have done that.

James:Do you think China is going to surpass the US in nominal GDP?

Helen: I think in nominal GDP, they probably will. Then there's this whole other question, I don't know how much of an international relations theory deep dive you want to do, but there's a whole other question of if we're thinking about China's rise in a holistic sense.

Obviously, GDP is just one measure and so there's lots of different ideas about if we want to talk about China rising to the level of great power that the United States is. People will give all kinds of different metrics. One that I think is interesting, which I think is due to Michael Beckley, is taking GDP and multiplying it by GDP per capita, the basic idea being that the total size of an economy matters but also the GDP per capita matters because that's about how much wealth each individual citizen has, how much spare money does the government have to use for things and so on and so forth.

And then, the GDP per capita starts getting all entangled with demographic factors, like: are China's fertility rates slowing too quickly? The Chinese goal is to get rich before they get old, meaning to reach a higher income level before the population is an aging population in the way that you see in Japan or even the US to some extent.

Anyway, lots of interesting questions there -- I'm not totally sure which bits can be picked apart to try and make a concrete prediction or bet.

Not yet a subscriber? Join 75,000+ curious readers who grow with us every day

No spam. No nonsense. Unsubscribe anytime.

Great! Check your inbox and click the link to confirm your subscription
Please enter a valid email address!

James: Well, which ones are most interesting to you? I love that stuff, by the way, it reminds me of OBPS, on-base plus slugging percentage in baseball, where they take these stats and they combine them together to give a fuller picture, I think that's really clever.

Where do you see the biggest gap? You're in Washington every day and you're having, hopefully, very sophisticated versions of this conversation. What do you think is the biggest gap between what you see on the front page of the paper versus what might be worth making a bet about?

Helen: Yeah, I think in Washington, there's lots of different opinions. So, I don't feel like there's an expert consensus here that is different from what you would see on the paper. To state it in simplified terms, the big debate would be is China just rising so the US is at, maybe, the peak of the relative power it's going to have for this foreseeable future? Or, is it more the case that China is currently rising but the wave is going to crest and fall in some sense and how could we tell and what did that depend on?

James: Do you think the wave is going to crest and fall? Or, yeah, probably better to say, when do you think the wave will crest and fall?

Helen: I don't know if we can take a Googling break to go check out some stats. My best guess is that I have a hard time seeing China continue to just eclipse the US more and more. I do think they're going to face some really significant challenges escaping what gets called the middle income trap which is, essentially, you see lots of countries that were poor, managed to industrialize somewhat but really struggled to get out of the middle income range where you don't have serious problems of extreme poverty but you're not at the industrialized wealthy nation level that you see in the OECD, for instance. So, my sense, having never actually dug into this that closely, is that I do expect China to start to struggle a little more in, say, the 2030s.

Helen: Yeah, and maybe the wave cresting and falling a little bit there but I've never actually really dug into the numbers to come up with a firm version of that overall sense.

James: Well, what better time to do it than live with the Browser? I think it's really interesting. What would drive the slowing down? What keeps China from just continuing on to become like, I don't know, Europe?

Bet 1: Can China Get Rich Before It Gets Old?

Helen: Yeah, so a huge issue is this whole thing of can you get rich before you get old? Meaning what proportion of your population is working in the labor force, producing economic value which, in the simplest form, can be used to socially support your elderly population. So, the really simple version of this is you have to collect taxes from people who are working and so, if you have a ton of people working and a very small age population, then it's very easy to use your taxes to support the elderly. But China in the '70s instituted the one child policy because they were very worried about excessive population growth.

And so, since then, fertility has really cratered deliberately, they did that to themselves. Now they're facing this demographic time bomb, which is what it gets called if the number of people who are in full-time work, who are producing that economic value that can create a tax base that can be used to support the elderly but, also, to support everything else, right?

You use your tax base to build your military or to fund education or technology development as well. So, you can look at different curves of the total working population or the ratio of working population to non-working population or things like that and my impression is that, sometime in the 2030s, that's likely to peak and start to give them problems.

James: As a relative normie who sometimes lurks in China-watching communities, I feel like the biggest thing that is known in China-watching communities that is unknown in the rest of the world is that demographic chart. I remember the first time I saw it, just being like, "Whoa, how is this not framing every thing ... " Because periodically, there's angst in the United States around the relative competition between US and China and the demographics for the US are quite favorable in comparison to China. China's are the worst around.

Helen: It's tough. And the US does way better than a lot of comparable countries because the immigration levels are so high. We just let in younger folks and that actually really helps, it has been a huge advantage for the US.

I should say predicting what's going to happen with the future of the Chinese economy has been a fun game that folks in Washington played for decades and don't necessarily have a great track record on, so it could well be that we're mistaken again. They're definitely very serious, thoughtful analysts who think it's very possible that the Chinese government is going to find ways to steer their economy carefully through this and end up with the growth that they need to emerge out the other side. So, I think that's definitely possible, it's just not, not necessarily what I would bet on.

James: Not your bet, yeah. Okay. So, I like the idea of betting on the middle income trap. How would we know if they got caught in the trap?

Helen: I think that would be the most straightforward way. We could also go into an hour's long discussion of the right methodology and metrics but I'm guessing that picking a straightforward existing stat is the way to go. Yeah, I'm pretty sure the World Bank has categories of low income, middle income, higher income. They might have more than three categories, I'm not sure but we could maybe look at what that count is, look at where China is now and see if there's a date by which it's predicted they'll exceed that or if we want to predict that they'll never exceed it.

James: Well, this is your prediction. So, what do you think?


Helen: Yeah, I want to go look at some trend lines and pick them. Is that allowed?

James: Yeah, go look at some trend lines.


[Googling break begins]

James: Well, let's bring this back in and we'll talk it through because I think this would be the interesting part. The Googling break is done, we've now figured out what data we want to use. Helen, walk us through how we're going to structure it and then we'll figure out where we stand on the bet.

Helen: Interestingly, it looks like China is right at the upper limit of what the World Bank would call a middle income country. They're maybe a thousand or $2,000 off and they're at something like $10-ish thousand per capita and the World Bank defines it as up to $12,000 or something. So they're looking pretty good.

The thing that we thought might be interesting would be to compare them to a country that is generally considered to have gotten rich before it got old, for instance, South Korea. And so eyeballing South Korea's trajectory here, it looks like they were where China is now in 1999 or so. And then, say, 10 years later, they were up at about 20,000 USD.

That was right after the financial crisis so the year after, they would jump up again. So, 20,000 may be a little over. So, maybe one thing we could do would be, “What's the likelihood that China is there in 10 years? So, in 2031?”  Or, actually, today we're looking at our 2020 data so maybe we should say 2030 data. In 2030, will China be above 20,000 US dollars per capita income?

James:I think it's a great question. What side do you fall on?

Helen: These predictions are always totally subjective, right? Intuitively, I feel like I would give them about a 50/50 chance which might sound like a cop out answer but I don't think it is. Because if we were saying like, "Oh, yeah, they're on the path of South Korea then they would have a really good chance of being up there." And I would expect them to have a harder time than South Korea, but I don't feel sure of that. I could definitely imagine them managing it. I would be about a coin toss that they're going to make it. What do you think?

James: So I'm going to be a little provocative here and I'm going to step outside of my area of expertise. One of the benefits I have is this is not what I work on. If you're going to make the positive case for China, you'd be like, listen, they're basically becoming the platform for manufacturing for the whole world and what they're going to do is find ways to get other countries to fit in to what they're doing and they're going to use that to get to the technical frontier in some areas and to use their size and their scope to defy demographic gravity a little bit. So, I'll take the over. I'll say in 2030, the China GDP per capita is above $20,000, I think, in [2020 US dollars].

James: It raises another question. Is that enough to tip the balance? What does that mean for world power? You probably have a better answer to this than I do.

Helen: Yeah, I actually don't. There's stuff on my reading list that would go into this that maybe I should have read before having this conversation with you but we're here now. What is often talked about is the middle income track. And so, if we’re saying middle income is 12 and a half thousand, it would seem like if they get to 20,000, they're out of the middle income trap.

On the other hand, if they hit 20,000 and that turns out to be the peak of their success, then maybe that still doesn't look so great for them because Korea, for instance, got to 20,000 in around 2009, 2010. Now, they're up at around 30,000 which is pretty decent though still not as good as Japan or Singapore, which are the nice comparison lines Google is offering me in this automatic chart. So, yeah, I feel like 20,000 would mean things are looking pretty promising, nothing has gone wrong yet for them. And then that would then set up the question of what happens in the 2030s.

Fences Around The Capitol

James: Okay, before we move on to our next topic, I do have a couple of questions for you. I understand that you have taken on a pretty public bet recently. Do you want to tell us a little bit about it?

Helen: I did. Yeah, I think it might be my first Twitter bet. A guy I follow does geopolitical consulting. Shortly after the January 6th insurrection at the Capitol, he tweeted, lightheartedly, about discussing with his friends and colleagues all of this fortification that has gone up around DC. There were lots of new fences, lots of armed guards standing around and they were talking about is this going to go away? When will it go away? And he was saying, "Consensus among me and my friends and colleagues was it's going to stay. Congress is going to decide that they really like this." And that didn't seem right to me based on my picture of how Congress feels about fortifications and also how DC residents feel. People in the Capitol really love sledding down the hill on the grass with their kids in the winter and things like that.

And so, I took his somewhat lighthearted but somewhat serious tweet and asked him if he would bet about it. And, maybe a nice story in the perils of betting, the challenges, which I'm sure you're very familiar with from running this series, is how do you get it concrete enough... We both knew we had to come up with something very concrete so we would know did this get met or not. What occurred to me was there's this nice plaza area on the western side of the Capitol where, in 2019, in the summer, I had been to jazz concerts and I knew that the public could get in there very freely. And so, my suggested way to operationalize the bet was, I bet that by the end of this year, so at some point during 2021, I will be able to walk onto that plaza without going through a gate or without going through any kind of checkpoint.

And I think you're grinning because you know the end of the story or not the end, it's not December yet, but my most recent time there a few weeks ago, it was still blocked off, that specific plaza. So, all of the fortifications are down, the fences are down, there's way fewer armed guards around and National Guard and so on are all gone. But there's blocks around that specific plaza. I called out to a security guard [there] saying, "Hey, do you have any idea if these are coming down soon or if they're going to be permanent?" And he said, "Oh, well, they're starting stone work very soon." And so, I don't know if they're going to stay up for security or for the stone work but probably it's going to stay closed. So, that's a little bit of a bummer.

On December 31st, when this is all done, we can go back and revisit. The specific wording of the bet I chose which was to be able to walk up to the Western steps of the Capitol, I think, actually leaves some room for interpretation but, in my heart, I know that I was thinking of walking onto this plaza. And so, I don't know if I'm going to be able to claim victory in that bet even though I think his ultimate hypothesis initially was wrong. But I'm the one who picked the terms, so…

James: If you're a DC stonemason out there working on this, we'd really appreciate a little hustle. As I understand your day job, you do a lot of thinking about how different tech trends are going to impact US national security and then provide different stakeholders throughout the government with ways to think about that. Is that a fair synthesis of it?

Helen: Yeah, I think that's right. I should definitely say I'm here in my personal capacity, just spitballing stuff for me. But that's a decent summary of what I do for work.

James: Excellent point. I guess I was curious to what degree you think of what you're providing as a prediction and are you ever asked to make specific predictions or falsifiable predictions? For instance, for me, it's pretty uncommon to be asked to make a falsifiable prediction but it's pretty common for me to be asked to give a forecast, right?

Helen: Yeah. I think we are pretty often being asked for some version of a forecast but usually not very explicitly. We actually have a dedicated forecasting platform called Foretell. I'm pretty sure the website is cset-foretell.com, which does crowdsource forecasts on the kinds of questions that we work on. And then for me, I think usually it's more like maybe the kind of thing you were describing. It's more like a question that you're asked is implicitly asking you to include some forecast but it's pretty rare to be asked straight out what percent likely do you think it is that such and such thing will happen or how do you think this trend will evolve in a very explicit forecasting way.

Bet 2: Automated Drive-Thru Orders

James: What do you want to bet on next? We already have a 2030 bet about China. AI?

Helen: Yeah, there's so much here. I'd love to hear input from you as well on whether there are things because, with AI, there's all kinds of different technologies. There's the research side, what kinds of things might we see published in the literature versus what might we see deployed in the real world. There's all kinds of interesting regulatory activity going on right now as well. The EU has its AI Act that's coming out. I think Congress in the US is trying to figure out if there's stuff it should be doing. I don't know if any of those feels particularly enticing to you.

James: So what is interesting to me about AI as a topic and I don't want to put any words in your mouth here, but I feel like it's an area that is generally overhyped. In my day job, I lived through the great AI rebranding where, all of a sudden, a software platform that I worked on went from being a big data platform to being an AI platform without any changes that I or anyone internally was aware of hitting the product.

Helen: That's great.

James: So, I guess I feel like, in terms of how day-to-day life is going to change, over the next five years, I feel like it's overhyped but I'm not totally sure how I would turn that into a bet and I'm hoping, with a little bit more domain knowledge, you might be able to. You might disagree but you might be able to say this is how we measure AI.

Helen: I suspect there is going to be good stuff that we can bet about here because I definitely have some skepticism about a lot of the hype. Certainly anyone selling products has slapped a big AI label on anything involving even a hint of data science. But I also think that there's some pretty cool research going on and some of that research actually gets implemented pretty quickly or gets into the real world pretty quickly. I should definitely flag my conflict of interest being on the board of OpenAI here. So, I think they've done some really cool work on large language models, I don't know if you follow that at all.

Maybe for anyone who's unfamiliar, the basic idea is you train a model to take some text and then predict what comes next, what word or phrase is most likely to occur next after what it's already seen. And this just turns out to be this, surprisingly, multipurpose way of doing things because you can use it to generate a news article, you can use it for a chatbot. If the chatbot is getting to the chat, what's happened in our chat so far and predicting what a helpful customer service agent might say next. You can use it for fiction, you can use it for role playing.

James: I wonder if you could use it for bets. I wonder if we need to have the first AI Browser Bets.

Helen: That sounds amazing. I think you could figure that out. Actually, a friend of mine recently hosted GPT-3 on his podcast, GPT-3 being OpenAI's latest big language model. He fed the interview questions and then had, an AI-generated voice read out the text that GPT-3 produced. So, I wonder if there's something here, something that I would expect to be more likely than you would. Maybe something about AI, about how these large language models are used? One thing that OpenAI has been working on lately is this code generation work. So, they're partnered with GitHub which is a major coding open source platform and people can use GPT-3 to help them. Programmers use it to help complete their code which is pretty neat. There's lots of concern about ways this could be misused. So, whether we're likely to see it used in disinformation campaigns or things like that.

James: Well, what do you think is the most likely way that AI is going to start to impact our lives that people who went through the great AI rebranding might not expect?

Helen: Great question. I want to dispute the premise and say AI is already affecting our lives. So, I think all this brouhaha about Facebook that's going on as we record this, with their engagement based algorithms and the whistleblowing of things that the company knows about, the pitfalls if you're optimizing for engagement, things that your algorithms will promote, that's all deep learning, machine learning, the stuff that gets called AI. I'm also curious for you if there are things that you see people implying the AI will be able to do and you're like, "That's never going to happen." Or, "That would be if it was really smart, it would be able to do that but that's decades away."

James: I'll admit that what I'm about to say is definitely taste. As somebody who had a familiarity with machine learning and neural networks before I was hearing AI as the term for them, back when that was big data or machine learning, I would say that I always thought of AI as a system that is learning for itself versus a system where we can feed in a bunch of data and get really, really good predictions.

So, I think about some of the things that Google has done with Google Go as being more in the AI camp because, at least, as I understand that work and I'm definitely not an expert here, but they allowed the system to play Go and it learned. I don't work at Facebook, what I perceive them doing with their newsfeed is they're throwing in a bunch of interaction data, event data and then putting together a model that they run in production that's specifically tuned for that use case.

So, I would definitely agree. Most of the last 10 years of my career have been focused on machine learning that's obviously affecting people's lives. I feel like we're not yet at the point where when I order food at a restaurant, it's being taken by a machine. Or self-driving cars you could argue about whether or not that matches the distinction that I gave but that's another great one that has been five years away for 10 years.

Helen: That's totally true. The one you just said, I wonder if this is what we could bet on, I don't know what I would predict. But an AI taking your order at a restaurant, I actually know the founder of a startup that got acquired by McDonald's to do voice recognition order comprehension for the drive thru. You go to the drive thru and you're having your order taken by an AI and you can order however you want, in natural language, just the way you normally would and it gets converted into an order for the cooks to work on.

So, I don't know if that seems to you like it would be something you weren't expecting. I don't know but I'm wondering. I don't think I have any proprietary information on that company. I'm pretty sure I've just seen what the founder posts publicly.

James: That sounds great. So, the vision is that there's going to be no person? Because the fallback here, what a lot of places end up doing is that it's AI until it breaks. And the game I end up playing on call centers all the time is how do I break the AI so that I can solve my problem.

Helen: Yes, totally.

James: But they're going to try to get rid of the person in the drive thru?

Helen: That's my understanding. There would still be a person. I think anything involving complicated robotics is going to take ages, that's just really hard. Because restaurants like McDonald's, they don't have a notepad where they're writing down your order. They have an order system that is the number and the items and it's all computerized. So, getting from you speaking to the speaker in the drive thru to that being in their system so that the people in the kitchen can start preparing, I think they're trying to get that to be fully automated. And I would feel like they have a pretty good shot at doing that in the next couple years but I guess it depends how high their error tolerance is and, yeah, I don't know. But that seems like pretty near term to me. That seems like something that is AI that is not far off.

James: And the franchisee adoption, I think.

Helen: Yeah, that's true, that's true. Yeah, we'll definitely have to pick. Are we talking one demo store or are we talking all McDonald's? Are we talking 50 stores across the country?

James: I think if either of us, within a driving distance or traveling distance, that's acceptable to us personally…

Helen: So, I'm going to go on a 24-hour road trip around the country to win this bet is what I'm hearing there.

James: Yeah. You're a busy person, you got a lot going on. But if either of us can go to a McDonald's and order with it going all the way to the kitchen without a person intervening in the next two years, I'll say you win and, otherwise, I win.

Helen: Yeah, I think I'll take that. Feels risky, I don't feel like it's a slam dunk bet but I think I'd take the over on that.

James: I love it. Okay. Oh, and I'll pay for your order if you win.

Helen: Likewise.

Big Data Versus Small Data For Bets

James: Awesome. Okay, before we move on to our next bet here, I actually have some questions about some of your work on AI. One thing I saw you write recently was about big data versus small data and AI. And I was wondering if you could talk to us a little bit about, basically, for bets, for predictions. Is big data better or is small data better?

Helen: For getting an AI model to make a good prediction?

James: Yeah, or tell me a little bit about how big data versus small data impacts AI prediction?

Helen: In most cases, more data is better. That's a pretty good rule of thumb. Usually, more data is going to get you better answers, more generalizable answers because the problem with a small data set is it might only cover a small part of the space. So, then, you're doing something else and the AI system has learned the wrong thing. But we did this work recently on small data and a big part of the reason for this was I think big data usually works better, but that's not to say you have to have big data to use quote-unquote AI or machine learning or even deep learning which is the current hottest, most advanced type of machine learning, it's being used for lots of things.

We wanted to do some work exposing for, especially policymakers and not people who are not working with machine learning day-to-day, saying that, "hang on”. Even though, as you were saying, big data has had this hype since the early 2000s or something…, actually, there's lots you can do with small datasets as well." It's less like there are use cases where you definitely want small data even if you have big data available and it's more like if you don't have big data, we're talking millions of data points or something, and you only have a couple thousand or a couple hundred data points, there's probably still some stuff you can do depending on your specific use case. We had a big paper come out and then a shorter article summarizing some of the ways that you can actually get away with pretty small datasets and still get good performance. And there is ongoing research to make those approaches better as well.

James: I love it. So, I think when we were talking earlier before the show, you've mentioned that you are from Australia and I actually was curious if you see any differences culturally between betting or predictions in Australia versus the US versus other places that you've lived?

Helen: Yeah, great question. I don't know. I feel like mostly Australia and the US are similar in that, betting, most people think of sports. In Australia, horse racing is pretty deep in the culture. The city I'm from, Melbourne, the Melbourne Cup is an annual event. It's called the celebration that stops the nation because we literally got a public holiday for this horse race.  S you have all your odds and you're betting. Maybe that means that people are a little more familiar with odds than they are here even though, personally, I find thinking in terms of odds very confusing for whatever reason.

I think the main place that would stand out would not be a country. I’ve spent time in Australia, the US, Germany and China…,  but it wouldn't be one of those countries but more like the San Francisco Bay Area, had a lot of people who are really excited to be thinking in terms of probabilities and percentages and bets. So I think that's where I picked up a lot of the habits of trying to take something that I disagree about with someone and trying to convert it into a bet. What I like about that is not just betting as fun but, also, that, often in the process of getting from a vague sense of I'm right and you're wrong, you get into a very concrete version of what are we actually talking about here. And I find that maybe half the time when I do that I actually end up agreeing with the other person or deciding that I was wrong in the first place.

James: So, that's where all great bets fall apart is in the terms?

Helen: Totally.

James: Well, let's move on to our third topic. I think we wanted to go back to China and talk about emerging technology. You're going to have to lead here because I haven't been to China since 2006.

Helen: Oh, boy. I haven't been to China since 2018, so that makes me already out of date by China's standards. Definitely a big theme in my day-to-day work is what technologies are going to matter in the future? Where are they going to be developed? By whom? So, I talk with a lot of people who want to know who's winning the AI race, whatever that means. But they essentially mean how is China doing in AI, how is the US doing in AI and then other upcoming technologies. Maybe one that could be fun, because I'm not at all an expert in it would be to talk about quantum computing.

I was just talking to a PhD student on the weekend at MIT, he was telling me how China's just doing really incredible stuff in quantum computing. That's definitely a technology that we get asked about, something that I just really don't have much expertise on. Other stuff, there's biotechnology somthere's always the ethical questions of, they cloned the first monkeys, which I think is not necessarily a technical advantage or technological advantage, but is more a difference of what's acceptable. They had the gene edited babies, where the scientist who did that got in trouble.  There's a range of things. Quantum computing is the one that's jumping out to me right now as potentially interesting.

James: Let's do it. I know very little about quantum computing which is usually what qualifies me to make a bet.

Helen: I love it.

James: What do you think about quantum computing?

Helen: Yeah, I also know very little so this could be embarrassing for both of us in hindsight. I guess we'll find out.

Some things that have happened in the last little while. So, quantum computing, I guess maybe for folks who are unfamiliar, it's a different paradigm of computing. So, the computing we have right now is called classical computing because it's based on classical physics. You probably know the computers use ones and zeros which, in electrical terms, is on or off. Is there current flowing here, is there no current flowing here and all of our computing is based around what can you build out of giant stacks of ones and zeros.

And quantum computing is, basically… whoo, I hope I don't mangle this. Basically, it's built on the idea that, instead of storing it in terms of one or zero, on or off, is there current flowing through this wire or not, you can store information in quantum states which are fuzzier and more overlapping. So, the highly simplified way to say it would be that it can be both one and zero at the same time. I think it might be more accurate to say it has pieces of one and pieces of zero all mixed up together or something.

Helen: And so, there's lots of theorizing about what you could do with these computers. But then, there's also this question of can we actually build a system that actually works like this? Because we know how to build a wire with current that runs through it or doesn't run through it, but building these little equivalents of…  a bit, if you have an on or an off.

So, we're talking about quantum bits, qubits, can you actually build a system that will store the quantum information that you need and I think, often, they need to be very, very cold or in vacuum environments to be incredibly stable so they don't get disturbed by outside influences of any kind. And so, I think recently, we've had demonstrations of individual qubits actually being built and actually being used to do things. I think last year, maybe 2020, maybe 2021, we had a couple demonstrations of what gets called quantum supremacy which is when you can do something with a quantum computer that is much, much faster than what you could do with a classical computer.

James: Sounds way creepier than what it actually is.

Helen: Yeah, quantum supremacy really does sound super creepy but it basically just means building a quantum computer that can do something better than a classical computer can.

So, that's where we're at right now. And then my newly gained knowledge from this PhD who I was just talking to is that, in terms of a physical qubit, the little thing that you have in the laboratory but they're very unstable and they're very unreliable. And so, what you really want is a logical qubit which is some number of physical qubits that you can configure together to be reliable enough that it is the equivalent of having one perfectly functioning qubit.:

And this PhD student who I was talking to was saying that, maybe, in the next three years or so, we might see the first logical qubit and that, when we do, everything will go completely insane. So, that was exciting. And he seemed to think that China was really well-positioned to do well there.

James: I love it.

Helen: So, it could be a timing based thing of when the first logical qubit or a location thing.

James: We could do both. We could have a tree bet here, this is great.

Helen: Oh, boy. But I really feel like I don't really know more than you do here.

James: Finally, we're on an even playing field. No, this is great. Yeah, how many years do you want to go out?

Helen: So, he was saying three. He works in this stuff which makes me think he might be overly optimistic. But it also might mean he's overly pessimistic because, sometimes, people who are in the lab every day are. No, but he's not in the lab, actually, I just remembered he was doing the theory. So, let's say our over enthusiastic PhD student said three, so maybe five is more of a reasonable place to be. Would I bet on there being a logical qubit in the next five years? Yeah, sure.

James: Okay, I'll take the other side of that one. I hate being in the place of betting against progress but this is where life has taken me. Geography, do you want to go-

Helen: What do you think? Do you have a leaning one way or the other? Is this something you followed at all, the quantum computing progress?

James: So, until recently, I worked at Google and I got an email at some point when Google claimed quantum supremacy about something. That team, I think, works in Santa Barbara and I remember thinking, "Wow, it'd be really nice to work for Google in Santa Barbara… Can I talk my way onto that team?” Then I quickly decide that that's not what I do.

I would like to say that this is going to be invented in the US because it fits with my aspirations for what our country is but I don't have any reason to think that. I have no facts base here other than it seems like Google and IBM have a lot of momentum. This is like headline James where I see headlines and I'm like, "Oh, yeah, it seems like there's some stuff going on there." I feel like if one major tech company is going to roll out quantum computing, it will probably be Google. It just seems like it's something they care about doing but that's all I got.

Helen: Yeah, yeah. I feel like I'm in a pretty similar position, honestly, which is ... I don't know, maybe I should be updating more based on this conversation with a PhD student about China being really serious here. My biases are definitely generally in the direction of believing in the strength of the US, especially on fundamental science and technology, it's just such a good track record and such a strong ecosystem. I don't know. But maybe I should go with China on this one if the whole reason I brought it up is that it's an area where they may be particularly strong and, if you're leaning towards the US, then maybe that's a bet we should make.

James: Let's do it. If you're an expert in quantum computing and would like to have an informed discussion on the Browser Bets, this is an open invitation, you can leave a comment below and we'll follow up with you.

Helen, before we close things off today, anything that you want to tell people about that you're working on or that you're interested in that you want to leave us with?

Helen: I don't know. I guess I would just say I do really think I have just got a ton of benefit out of trying to think more in terms of bets and not just thinking of it as a jokey thing that I do to win a few dollars off my friends. I genuinely do this with my husband where it'll be like we disagree about something and we'll say, "How can we make a bet out of this?" Because if we disagree, there must be some bet in there somewhere. Especially with him, the number of times that either we figure out the bet and we're like, "Oh, no, we actually agree on that." Or, we figure out the bet and he's like, "Yeah, I'm definitely betting with you on that." And I'm like, "Oh, wait, hang on. Now I'm not so sure. You're so confident then, probably you're right, I guess I'll just agree with you." So, if you have any friends who are up for having those conversations with you, I just want to say that I've really found them to be super great.

James: Yeah, I think what I've enjoyed as I've been thinking about this series is all the different places and cultures where betting comes up, my mom, growing up, all the time would be like, "Well, let's bet a quarter on it,". It would be like would we make it to church on time.  And this has followed me throughout life. Different groups that don't have a whole lot in common still have this ritual of, "Okay. Well, let's have a bet."

Well, thank you, Helen, for taking the time with us today and I look forward to getting an AI milkshake with you at some point in the next two years.

Helen: Sounds fantastic. I look forward to us both somehow managing to be wrong about the quantum bets. I don't know how but it seems like what will happen.

James: It'll be tough to be both of us wrong on the timing but I'm sure there'll be a way. All right, thanks Helen.

Helen: Thanks a lot.



Comments

Join 90,000+ curious readers who grow with us every day

No spam. No nonsense. Unsubscribe anytime.

Great! Check your inbox and click the link to confirm your subscription
Please enter a valid email address!
You've successfully subscribed to The Browser
Welcome back! You've successfully signed in
Could not sign in! Login link expired. Click here to retry
Cookies must be enabled in your browser to sign in
search