A sad, nesto sto se obecava, vec duze vreme:
Progress towards fully capable and practical
quantum computers isn't slowing down, and researchers from Google are the latest to announce a significant step forward in the capabilities of today's machines.
While we call these devices
quantum computers, they're
more like prototypes of what quantum computers can be: At present they require super-specific, extreme conditions to operate in, and struggle to stay stable and error-free.
Despite those limitations, their computing potential is becoming more impressive all the time.
The latest system
run by Google has a total of 70 operational qubits – the quantum equivalents of classical bits that can represent 1, or 0, or both at the same time, potentially allowing for certain calculations to be performed at astonishing speeds.
Specifically, the team used a complex, synthetic benchmark called random circuit sampling, which is
exactly what it sounds like – taking readings from randomly generated quantum processes.
This maximizes the speed of critical actions, reducing the risk of outside noise destroying the calculation. They then estimated how long it would take existing supercomputers to run the same sums.
"We conclude that our demonstration is firmly in the regime of beyond-classical quantum computation,"
write the researchers in their recent paper.
The
Frontier supercomputer, currently the most powerful computer in the world, would take a little over 47 years to crunch the same numbers, the researchers suggest, whereas the
Sycamore quantum computer managed it in mere seconds.
A group including Google engineers
did something similar in 2019, with 53 qubits. Then, as now, there's a debate to be had about how useful and practical these particular simulations are, and how fair (or otherwise) it is to compare supercomputer performance to what has been managed here.
Nevertheless, the Google team is clear in its claims that this demonstrates
quantum supremacy: the idea that quantum computers really can deal with processes above and beyond anything that even the fastest classical computers can cope with.
https://www.sciencealert.com/google-quantum-computer-is-47-years-faster-than-1-supercomputer
Prostim jezikom covek ima 3 komponente: potvrdjivanje, negiranje i odredjivanje sta od ta dva.
Tjuringova ima samo prva dva i zato je AI nemoguc.
No kvantni imaju sva tri, dok je rasudjivanje stavjeno na random, pa ne daju predvidljive rezultate.
To je isto kao kada bi sve odluke donosio bacanjem novcica, rade na tom problemu...
