Now, a panel asking “What will quantum computers be good for?” Thoughts? What will quantum computers be useful for? And when? #FQXi2019 
Scott Aaronson: Uses are more specialised than popular articles suggest. But amazing there are applications at all: (1) simulating chemistry and physics (Feynman's original idea in '80s). Useful for material science, designing high-temp superconductors, better batteries...
Aaronson: ...use of quantum computers: (2) Breaking cryptography -- all systems used to secure the internet, but we can switch to more secure systems (3) Grover's algorithm speeding up searches, (4) quantum machine learning -- but hard to show better than classical equivalent...
...(5) I proposed a protocol for a q computer to produce random bits and then prove that it is truly random. So use quantum computers as "provers".
Aaronson: But we don't yet know the limits of what classical computers can do.
Michele Reilly: Agree, quantum computers could be great for simulations. Some exciting ideas from Daniel Harlow & others. Need to reduce resources for these. Sachdev looking at dualities. Quantum gravity.
Michele Reilly: Google has 72 qubit architecture, costs 100k per chip. But you have to also take into account the classical peripheral overheads (fridges, cables, etc) all adds up a million dollars. And we'd need 10^6 qubits. Would cost more than the LHC. Cost/benefit issues.
Mile Gu: We all agree on basic applications for quantum computers. To be more controversial, let's think in the distant future and could overcome monetary and resource constraints, what other applications could we have?
Gu: Could address the measurement problem of quantum physics (why observation collapses a quantum system)? Can we reverse collapse in the future. (Paper by Calvacanti: arxiv.org/abs/1907.05607)
Gu: Near-term application for quantum systems. We know classical information needed to track a quantum system can grow unbounded. May require unbounded classical memory.
Seth Lloyd: Google, IBM, etc... we're at a really exciting time in terms of what's being done now. Can investigate Many-Body Localization problems fairly soon, after a few quantum engineering issues are solved. D-Wave, whatever you think of it, could be good for this.
Gu: Classical computers scale linearly. Interesting thing about quantum computers, complexity seems to grow exponentially. Interesting social and political consequences. Incentive for countries to work together to do something far greater than what they can do themselves.
Lloyd: Even people doing AdS/CFT and string theory are doing quantum information theory. "There goes the neighbourhood."
Comment from @FrancesVidotto: would make a fantastic simulator for loopy spacetime.
Question from @tegmark: Can you use a really large quantum computer to rule out quantum interpretations? Aaronson: That's not a quantum interpretation, but a rival theory that makes different predictions to standard theory, which might prevent a q computer being built. eg GRW
Lloyd: GRW already bounded to such an extent, you could treat this as noise with quantum error correction and still build a quantum computer.
Question from Matt Leifer: Being too optimistic! What is the most destabilising outcome that you can think of?
Lloyd: Most dangerous thing about quantum computers at the moment is if you drop one on your foot.
Aaronson: Most dangerous applications of a quantum computer? Obviously Shor's algorithm leading to someone hacking someone's emails, affecting the outcome of an election and leading to some clown getting elected. Can't imagine that happening! #FQXi2019
Gu: Can't see a catastrophically destructive outcome.
Question from @SimonDeDeo: could a quantum computer be used to prevent something else in the lab from decohering, using quantum error correction. Aaronson: I think the other thing would also be a quantum computer.
Question: The best classical algorithms that have yet to be found could be better than quantum algorithms. Could quantum computing be successful even if no q computer has been built, but it has driven great improvements in classical computing.
Aaronson: Sure, depends how good they are. Like asking is string theory a success if all it does is teach us more about condensed matter physics theories? Sure, but depends how good the theories are.
