3 ms·

Before LLMs/AI became the obvious "next big thing in computing" I remember coming across a fair number of opportunistic devs on LinkedIn trying to promote thems

Before LLMs/AI became the obvious "next big thing in computing" I remember coming across a fair number of opportunistic devs on LinkedIn trying to promote themselves as "quantum software engineers", and even a just a year or two ago I would see "quantum machine learning" on people's profiles. I remember thinking maybe I had missed something and seeing how many qubits we could even have... needless to say it was (and still is) not enough for any meaningful ML work quite yet. If you search you can still find some, and, as someone who has spent more than a decade doing actual machine learning, I find the audacity to claim that you're doing any kind of serious software engineering, let along proper ML work, on a quantum computer to be almost impressively audacious.

amitav1 11mo ago
As a fellow currently dipping my toes into quantum machine learning, I think that you think we're saying "machine learning on quantum hardware", when what we actually mean is "machine learning for quantum computing on classical hardware". That is, using machine learning on classical computers to try to increase the effectiveness of quantum hardware.
- hulitu 11mo ago
  > what we actually mean is "machine learning for quantum computing on classical hardware" No blockchain ? what happened to blockchain ? /s
volemo 11mo ago
I know a couple of quantum software engineers and these people are in universities writing novel algorithms on whiteboards (and sometimes testing them out in QuPy).
- georgeecollins 11mo ago
  As I understand it Quantum algorithms are very much needed as the hardware improves. The hardware gets shown off, but there aren't a lot of algorithms that can take advantage of it if it worked better. Yet.
  - volemo 11mo ago
    As I understand it, the hardware very much lags behind the algorithms. We have plenty of cool algorithms to run on quantum computers, but to be practically interesting they all need more qubits or better coherence times than is available today.
    - hershkumar 11mo ago
      The hardware very much lags behind the algorithmic advances, much of the current push for new features in quantum hardware (midcircuit measurement/feedforward, phonon mode coupling, etc) often comes from theorist colleagues pestering experimentalists about whether their hardware can run their algorithms yet. In fact, this is analogous to the original motivation for the development of classical supercomputers, physicists wanted to run expensive non-perturbation Lattice QCD calculations, so they co-designed some of the earliest supercomputer architectures.

amitav1 11mo ago

As a fellow currently dipping my toes into quantum machine learning, I think that you think we're saying "machine learning on quantum hardware", when what we actually mean is "machine learning for quantum computing on classical hardware". That is, using machine learning on classical computers to try to increase the effectiveness of quantum hardware.

hulitu 11mo ago
> what we actually mean is "machine learning for quantum computing on classical hardware" No blockchain ? what happened to blockchain ? /s

volemo 11mo ago

I know a couple of quantum software engineers and these people are in universities writing novel algorithms on whiteboards (and sometimes testing them out in QuPy).

georgeecollins 11mo ago
As I understand it Quantum algorithms are very much needed as the hardware improves. The hardware gets shown off, but there aren't a lot of algorithms that can take advantage of it if it worked better. Yet.
- volemo 11mo ago
  As I understand it, the hardware very much lags behind the algorithms. We have plenty of cool algorithms to run on quantum computers, but to be practically interesting they all need more qubits or better coherence times than is available today.
  - hershkumar 11mo ago
    The hardware very much lags behind the algorithmic advances, much of the current push for new features in quantum hardware (midcircuit measurement/feedforward, phonon mode coupling, etc) often comes from theorist colleagues pestering experimentalists about whether their hardware can run their algorithms yet. In fact, this is analogous to the original motivation for the development of classical supercomputers, physicists wanted to run expensive non-perturbation Lattice QCD calculations, so they co-designed some of the earliest supercomputer architectures.