Google talks about quantum supremacy: quantum computing and AI can promote each other

Google talks about quantum supremacy: quantum computing and AI can promote each other

In a paper published in the British journal Nature on the 23rd, Google demonstrated quantum supremacy — a programmable quantum computer that surpassed the fastest classical supercomputer. The quantum system took only about 200 seconds to do what a classical computer would have done in about 10,000 years. The review article called the achievement a major milestone in quantum computing.

One goal of quantum computing is to perform certain computational tasks exponentially faster than conventional classical computers. Achieving this goal requires overcoming many challenges, such as generating a large computational space while keeping the computational error rate low, and designing a benchmark that is difficult for classical computers but can be easily accomplished by quantum computers.

The research team of Google AI Quantum and the University of California, Santa Barbara, described the technological progress made in achieving quantum supremacy. They developed a 54-qubit processor that uses quantum superposition and entanglement to achieve an exponential increase in computational space compared to what can be achieved with classical bits. With 1 qubit not working efficiently, the processor actually uses only 53 qubits.

The error correction process developed by the research team can guarantee high operational fidelity (up to 99.99%). To test the system, the team devised a task to sample random numbers produced by quantum circuits. For a classical computer, the difficulty of this task increases as the number of qubits in a quantum circuit increases. In the end, the quantum processor took 1 million samples from the quantum circuit in around 200 seconds, a task that would take a cutting-edge supercomputer about 10,000 years to complete.

In a concurrent “News & Views” article, the scientists commented that “achieving quantum supremacy for the world’s leading supercomputer…is undoubtedly a remarkable achievement”. But the article also points out that more work is needed before quantum computers can be put into practical use, such as achieving sustainable fault-tolerant computing.

According to foreign media reports, in a paper recently published in the journal “Nature”, Google researchers claimed to have achieved “quantum supremacy” (also known as “quantum supremacy”) for the first time. Their 53-bit quantum computer, called Sycamore, took 200 seconds to complete a computational task that, according to Google, would take the world’s fastest supercomputer 10,000 years. (Parts of the paper were disclosed and reported online last month)

This time the calculation has little practical use – it spits out a string of random numbers. It was chosen simply to show that Sycamore can indeed work like a quantum computer. Practical quantum computers are still many years away, the technical hurdles are huge, and even then they may only beat conventional computers at some tasks.

But Google’s research is still an important milestone — Google CEO Pichai likened it to the Wright brothers’ maiden flight in 12 seconds.

For a project that could take 10 years or more to pay off, outsiders wonder why Google has taken 13 years.

The following is an interview transcript of a foreign media.

Media: You have a quantum computer for a very narrow, specific task. What is needed to demonstrate quantum supremacy more broadly?

Pichai: You need to build a fault-tolerant quantum computer with more qubits, so you can apply it on a wider scale, perform computational tasks for longer periods of time, and thus be able to run more complex algorithms. But you know, if you want to break through in any area, you need to start somewhere. To borrow an analogy – the Wright brothers. The first plane only flew for 12 seconds, and there was no practical application, but it showed the possibility of the plane being able to fly.

Media: Many companies have quantum computers. IBM, for example, has a whole bunch of these facilities online that people can use in the cloud. Why can’t their machines do what Google does?

Pichai: The main thing I’m going to comment on is why the team at Google was able to do this. This requires a lot of systems engineering – working together on multiple aspects of the task.

From a systems engineering point of view, this is very complex. You actually start with a wafer and have a team actually etch the gate, make the gate, and then process the stack all the way to be able to use artificial intelligence to simulate and understand the best results.

Media: The last sentence of the Google research paper says, “We are only one creative algorithm away from a valuable near-term application.” What does this application refer to?

Pichai: What’s really exciting about quantum is that the universe fundamentally works in a quantum way, so you’re going to be able to understand nature better. It’s still early, but the highlight of quantum mechanics is the ability to simulate molecules and molecular processes, and I think that’s where it’s strongest. Drug discovery is a good example. Or fertilizer production – the Haber process produces 2% of the world’s carbon emissions, and in nature, essentially, the same process is done more efficiently.

Media: So how far do you think applications like improving the Haber process might go?

Pichai: I think ten years from now. We’re still a few years away from scaling and building quantum computers that perform well enough. Other potential applications for this technology could include designing better batteries. Either way, you’re dealing with chemistry. A better understanding of this technology is also where we are investing.

Media: Some say quantum computers may be like nuclear fusion: there may be breakthroughs in just the next 50 years. This seems like an esoteric research project. Why is the Google CEO so excited about this?

Pichai: Google wouldn’t be here today if it weren’t for the growth we’ve seen in computing over the years. Moore’s Law has allowed us to scale computing power to serve billions of users of a vast array of products.

So in essence, we consider ourselves a deep computer science company. Moore’s Law is at the end of its cycle, and quantum computing is one of many factors by which we will continue to make progress in computing.

Another reason we’re excited about this technology — take a simple molecule, caffeine (which has 24 atoms) has an infinite number of atomic states. We know that we can’t even understand the basic structure of molecules using classical calculations. So when I look at climate change, when I look at medicine, that’s why I believe quantum computing will one day drive progress in those fields.

Media: In 2012, when you saw a picture of an artificial intelligence learning to recognize cats on its own, you described in a media that you had a “hunch” that “this thing will zoom in and maybe reveal how the universe works, which will is the most important thing we do as humans.” Does quantum computing feel just as important?

Pichai: Exactly. Being able to actually manipulate qubits in the lab and put them in superposition was also a big moment for me, because in my previous opinion, that’s how nature works. It opens up a whole new realm of possibilities that didn’t exist until today.

Media: It may take a long time to get to a quantum system capable of doing something serious. How do you manage patience in a company accustomed to rapid growth?

Pichai: You know, I’m with Hartmut Nevin, who leads the quantum team with chief hardware scientist John Martini. I mentioned that I dropped out of my PhD in Materials Science. I used to work on high temperature superconductors and it was 26 years ago and I was sitting in the lab and I thought, “Wow, this takes a lot of patience to do.” I don’t think I have that patience. I have a lot of respect for the people on the team who have persevered on this journey for a long time. But almost all foundational breakthroughs work this way, and you need that long-term vision to build it.

The reason I’m excited about such a milestone is that while things take a long time, it’s these milestones that are driving the field forward. It was 1997 when the Deep Blue supercomputer beat Garry Kasparov. Fast forward to 2016, AlphaGo beat Lee Sedol at the Go game, and you can look at it and say, “Wow, that’s a long time.” But every milestone rewards those who do the work and attracts a new generation people enter this field. This is how human beings progress.

In terms of my previous systems engineering point of view – we’re pushing many layers of the quantum computing stack. So we’re pushing progress, and that’s going to be used in many different ways. For example, we build our own data centers, which allows us to build things like TPUs (Tensor Processing Units, which are specialized chips for Google’s deep learning framework, TensorFlow), which make our algorithms run faster. So it’s a virtuous circle.

One of the great things about working in innovative technology R&D is that you are valuable even if you fail, and even mid-term milestones have other applications. So yes, you are right, we have to be patient. But there was a lot of real satisfaction along the way.

Media: How much are you currently investing in quantum computing?

Pichai: It’s a relatively small team. But it builds on all the investments we’ve made at every level of Google over the years. It builds on years of company research and the applied work we’ve done on top of it.

Media: Can you talk about the difference in approach between Google and IBM? First, IBM has a bunch of quantum machines that are put in the cloud for people to program, and you do it as an internal research project.

Pichai: IBM is offering it as a cloud appliance, which is great and attracts other developers. I think we, as a team, have been committed to making sure we demonstrate to ourselves and the community that you can cross this important milestone of quantum supremacy.

Media: IBM also said that the term “quantum superiority” is misleading because it means that quantum computers will eventually do better than conventional computers, when in fact they have to work together on different aspects of the problem. They accused Google of exaggerating.

Pichai: My answer to this is that it is a technical term for art. People in tech circles fully understand the significance of this milestone.

Media: But the point of contention is that the public might see this as a sign that quantum computers have triumphed over conventional computers.

Pichai: I mean, it’s no different than we celebrate artificial intelligence. Some people confuse it with ordinary artificial intelligence. That’s why I think it’s important that we publish papers on quantum supremacy.

It’s important that those who explain these things help the public understand where we are and how you’re going to apply traditional computing to most of the problems you need in the world. The use of conventional computers will remain true in the future.

Media: AI is creating business for Google on many levels, and it’s making its way into services like machine translation and web search. You deliver AI tools to people through your cloud. You provide TensorFlow, an artificial intelligence framework that allows people to build their own tools. You also provide specialized chips that people can use to run their tools. Do you think quantum computing will eventually permeate Google’s business?

Pichai: Absolutely. If you take a step back, we invest in AI and develop AI, but then we know it’s useful to all of our businesses.

And finally, in all the practical applications you talked about, we don’t just use AI technology for ourselves, we deliver it to customers around the world, we care about the proliferation of AI technology, and the same goes for quantum computing.

Media: What do you think quantum computing means for artificial intelligence itself? For example, if you combine quantum computing and artificial intelligence, can it help us solve the obstacles of artificial intelligence?

Pichai: I think it’s going to be a very powerful symbiotic thing. Both areas are in the early stages of research. AI has exciting research topics in building larger models, more general models, and what computing resources are required. I think AI can accelerate quantum computing, and quantum computing can accelerate artificial intelligence, and I think that’s what we need to finally solve some of the toughest problems we face, like climate change.

Media: You mentioned the popularization of new technologies. Google has faced some ethical controversies around AI — who should use these tools and how. What have you learned from dealing with these issues and how has it influenced your thinking about quantum technology?

Pichai: Publishing papers and engaging with academia at these stages is very important. We work hard to participate. We have published our comprehensive AI principles. If you take an area like AI bias, I think we’ve published over 75 research papers over the past few years. So, we straighten out our ethics and actively participate in research.

I think there are areas where regulation might make sense. We want to engage constructively to help get the regulations right. Finally, there is a process of external engagement and feedback. These are all technologies that affect society. No company can figure out what’s right. There is no silver bullet, but it’s too early and we have to work together on all of these issues for the next 10 years.

Media: On the one hand, isn’t it a bit of a contradiction to say that you won’t develop AI technology for certain purposes based on AI principles, and on the other hand, to create a platform where people can use AI for whatever purpose they want?

Pichai: AI safety is one of our most important ethical principles. You want to build and test security systems. This is inherent in our development process. If you are concerned that quantum systems will break cryptography over time, you need to develop better quantum encryption techniques. When we built search engines in the past, we also had to deal with spam.

The stakes are obviously higher with these new technologies, but on the one hand is the technical approach you take, and on the other is global governance and ethical protocols. You will need to achieve the global framework that produces the results we want. We’re committed to doing what we can to help develop this technology, not only responsibly, but using it for security, etc. We will do this with other agencies.

Media: Are there other technologies that you are also very interested in right now?

Pichai: For me personally, there is a lot of potential for better ways to generate clean, renewable energy. But I’m very excited about the combination of all these technologies and how we can actually apply them. In healthcare, I think in the next decade or so we’re on the verge of a breakthrough that’s going to be profound. But I’d also say AI itself – the next generation of AI breakthroughs, new algorithms, better generalization models, transfer learning, etc., are all equally exciting to me.

The Links:   HR215WU1-120 EL640400-CB1 TFT-LCD

micohuang