The UCL quantum machine learning paper from April showed practical quantum advantage on chaos prediction. Does this change your view of the timeline?

[Sega26]

Wang, Xue, Gao and Coveney published in Science Advances on April 17. They demonstrated a quantum-informed ML framework that improved turbulence prediction accuracy by 17 percent and spectrum fidelity by 29 percent over classical baselines, with the quantum component running on a real superconducting processor. The code is open source on GitHub.

Is this the kind of result that should shift timeline expectations or is it too narrow a domain to generalise from?

[BackRowBob]

It is a genuine result. The turbulence application is commercially relevant and the quantum component ran on real hardware. That combination is rarer than the press release version of quantum advantage

[Storm52]

The domain specificity is important. Fluid dynamics is a good domain for quantum advantage because the underlying maths maps well onto quantum representations. It does not generalise freely

[BiscuitTin46]

The memory compression result is the one I keep returning to. Kilobyte scale Q-Prior from megabyte data is a different category of advantage from raw compute speed

[TommyB_20]

Anyone who has reproduced the result. The code being public is the right move. Independent reproduction is the only thing that matters

[DeepPilot]

Someone on this forum mentioned running the simpler test cases from the code last week and it worked. That is more than most quantum advantage papers can claim

[IronFist56]

The timeline this shifts is the fluid dynamics and materials simulation timeline not the general purpose quantum computing timeline. Important distinction

[Mike80]

Weather forecasting accuracy improvements in the 17 percent range have real economic value. This is not a benchmark designed to be impressive. It is a useful domain

[Zero-Point]

I remain sceptical that the quantum advantage survives full accounting of the classical resources used alongside the quantum hardware. The supplementary methods need more independent scrutiny

[Hollow Coder]

The honest shift in my view is that hybrid quantum-classical approaches are going to deliver value in narrow domains years before universal fault-tolerant quantum computers exist

[Pilgrim]

Not sure I am fully with you on that one. Happens to me all the time.

Ha, fair enough.