Preparing for Quantum. A Conversation with Scott Aaronson
We are on the cusp of a new tech global era.
It is no longer good enough to look around corners. We need to look around corners of corners. We need to see the potential before us, and be prepared—to take on new directions, new challenges and new unknowns. We set the stage for the next decade and beyond, with what we build today.
Quantum is one key driving force.
The new tech global era will enable climate innovation, electrification, digital currency, voice/facial recognition and extended reality. Scalable quantum, in fact, could become a reality within our professional lifetimes. I believe it will.
One person who is deep in the tech frontier is Scott Aaronson, Founding Director of the Quantum Information Center at the University of Texas, Austin, and AI Safety Researcher at OpenAI. In a recent episode of our OpenTalk speaker series, Scott shared some startling insights about the future of quantum and AI. Here’s a glimpse at our conversation.
Scott is an amazing thinker and leading expert, and it was deeply insightful spending time together.
Mark: Would you say there are quantum machines out there today that are working?
Scott: Yes, there absolutely are. It’s just that they’re very small ones. In the 1990s there were skeptics who said, “This is just ridiculous. You will never actually build this, because in real life all quantum systems are subject to power and noise.” Superposition states are very unstable. Any kind of interaction with the environment can collapse superposition.
A huge discovery in the mid- to late-90s that really convinced people that this can actually be done was something called quantum error correction: you don’t have to get the rate of leakage of your qubits into the environment all the way down to zero. You merely have to make it very, very, very small.
The goal has been to build qubits that you can act on accurately enough that then these error-correcting codes can get you the rest of the way. Then you want to be able to scale up to as many qubits as you like, maintaining their quantum state for as long as you need them to.
We are not there yet. All of the quantum computations that we can do, you could say are more or less impressive circus acts. Everything is going to fall apart after some number of steps, but you can try to make that number as large as you can. The state of the art today is systems with a few dozen qubits.
Mark: Humans are fallible, and we’re taught that computers are deterministic, at least in Von Neumann architecture. Do computers have to be deterministic? Can they be fallible?
Scott: Computers can certainly be fallible. Anyone who has tried ChatGPT over the last months has seen this! We now have these incredible AIs. You can ask them to prove that there are only a finite number of prime numbers, and they will happily oblige you with a proof that looks superficially plausible, but of course has some freshmen-level error in it, because the statement it’s trying to prove is false.
So, you can say, in a certain tautological sense, a computer always does what the laws of physics say that it would do. In that sense, it never makes an error. But in the humanly-relevant sense, of course they can be fallible, and we have daily experience with that.
And a quantum computer is no different. The one real difference with a quantum computer is they are inherently probabilistic. The whole point of the quantum computer is to create this superposition, this vector of amplitudes, and then make a measurement that converts those amplitudes into probabilities. So, given that everything is probabilistic, what do we even mean by a quantum computer succeeding?
This was one of the early questions that people like Umesh Vazirani had to think about when they invented the mathematical foundations of quantum computing about 30 years ago. And what they said was simply, “We will define a quantum algorithm for a problem to be a good one if we can make the probability of an error to be as small as we would like.”
Mark: You mentioned this earlier, and I still don’t understand what it means: a negative 30% chance of rain. Is that just a stronger zero, or does it mean something different?
Scott: A negative 30% chance doesn’t mean anything. It is every bit as nonsensical as it sounds! The whole framework of probability theory only really works with numbers from 0 to 1. But that is why it is so surprising that in quantum mechanics we have to use these other numbers called amplitudes, which can be negative or even complex. Now the key point is the amplitudes are not probabilities. They’re sort of pre-probability, the fundamental numbers that nature keeps track of. And then they get converted into probabilities when we actually make a measurement at the end.
Will ChatGPT pass the Turing Test?
Mark: It feels like during the pandemic of the last two to three years, a decade of progress has been made. Over the next 10 years, do you think we’ll make a hundred years of progress, and in your world view, where do you see that progress happening over the next five, ten years?
Scott: Well, progress is tricky because sometimes it goes ridiculously fast or faster than people expected in certain areas, while also going slower than they expected in other areas. If you had asked someone in 1970, they might imagine that by now everyone would have flying cars, that we would have space elevators, that we would have all kinds of things that we don’t have.
Mark: Quantum teleportation!
Scott: Right! But then they might be pretty amazed that we all carry these devices in our pockets that have instant access to the whole world’s information. And that might go even beyond what they fantasized about in their science fiction.
So, it’s really hard to predict in what areas the progress will be. But I think the next decade is going to be an utterly insane time for AI. I hope that it will be for quantum computing also. I hope that we’ll build a quantum computer before we just build an AI that can build the quantum computer and everything else for us.
Mark: Well, let’s be ready for Q2K when we get there!
I am an optimist about the transformative power of information and digitalization. Whether you’re a quantum skeptic or a quantum enthusiast, quantum is going to have a massive impact on Business 2030. This is especially true, as Scott argues, if quantum computers do not work as predicted.
It is time to prepare for Quantum.
Keep watching this space for more insights from “OpenTalk with Mark J. Barrenechea,” my conversations with some of the world’s greatest thinkers and leaders.