Quantum computer machine learning advantage over classical computers is fake news.

Home page
Quantum computer is useless

The recent research on "quantum computer machine learning outperformed classical computing in real weather forecast" is fake.  ← A classical computer easily outperformed this fake quantum computer.

(Fig.1)  Quantum machine learning with just 9 useless atomic nuclear spins cannot outperform a classical computer, contrary to fake news.

Fake quantum computer's machine learning.

Fake quantum computer consisting of only 9 useless nuclear spins can neither calculate anything nor outperform a classical computer, contrary to overhyped fake news.

Overhyped quantum computers still cannot factor even 21, much less conduct calculations or machine-learning.

So the recent research on "quantum computer machine learning system might outperform classical computing AI methods in real weather forecasting" is clearly overhyped fake news (= in this research, a classical computer easily outperformed this fake quantum computer's system consisting of only nine atomic nuclear spins ). ↓

Just 9 atomic nuclear spins are still Not a quantum computer nor outperforming a classical computer, contrary to hyped news.

This overhyped news (4/2026) on quantum machine learning advantage ↓

4th-paragraph says  -- Fake quantum advantage
"This represents the first experimental demonstration of quantum machine learning outperforming large-scale classical models (= which doesn't mean quantum computers outperforming classical computers ) on real-world tasks (= false )"

11th-paragraph says  -- No quantum computers
"To build this, the team used nuclear magnetic resonance techniques (= irrelevant to quantum mechanics, this-p.1~p.2 ) to control nine atomic spins (= Just 9 atoms are still Not a quantum computer ) ... When input data is encoded into this system,"

2nd-last-paragraph says  -- Useless quantum computer
"That said, the approach is still in its early stages. The current system is limited in size and has only been tested on specific types of problems.... scaling it up will bring new challenges (= still useless quantum computer's machine learning )."

This research's fake quantum machine-learning relies on a classical computer's training, and a classical computer numerical simulation easily outperformed the fake quantum computer of only 9 nuclear spins, so No quantum advantage.

↑ This research paper ( this-last-link ) this ↓

p.1-right-1st-paragraph says  -- No quantum machine learning
"QRC (= quantum reservoir computing = useless quantum computer's machine learning ) has so far been explored primarily through theoretical studies and numerical simulations (= classical computer simulation ) on various idealized quantum models, while experimental realizations remain relatively limited"  ← Still No experimental realization of actual quantum computer

p.1-right-1st-paragraph says  -- Error-prone quantum computer
"However, such ( quantum ) circuits often demand substantial depth, making them vulnerable to gate errors and dissipation, thereby limiting the fidelity and performance of the QRC (= quantum machine learning ) in the experiments"

p.2-left-Model design says  -- Classical computer training needed
"the input s_k and the target output y_k,.. The goal of QRC is to approximate this relationship by generating an output sequence y_k that closely matches the target"

"To extract the information, specific observables are measured at each time step, yielding a sequence of readout vectors x_k (= directly-measured results of 9-atomic quantum computer from input parameters )... The output is computed as a weighted sum of readout components, y_k = x_kw"

"the readout weights w are optimized on a training dataset by minimizing the error"  ← These weights w values to reduce errors was trained by an ordinary classical computer and datasets, so No quantum computer's machine learning.

p.3-right-2nd-paragraph says  -- just 9 nuclear spins, Not computers
"The quantum reservoir is a 9-spin coupling system consisting of four carbon nuclei and five proton nuclei in a 13C-labeled crotonic acid molecule...its scalar input is normalized to s_k and encoded via an x-axis radio-frequency pulse on proton nuclei,  ← Just controlling 9 atomic nuclear spins by radio waves modulated by input parameters.  ← Still Not a quantum computer "

p.13-last-paragraph says  -- Classical numerical simulation
"To validate the reliability of our experimental results, we performed numerical (= classical computer ) simulations of the nine-spin quantum reservoir computing model on the NARMA task,"

p.14-2nd-paragraph says  -- Classical (= numerical ) beat quantum
"The simulation results, presented in Fig. 7(a), are qualitatively consistent with the experimental results in Fig. 3 of the main text. Quantitative differences are attributed to deviations between the simulated dynamics and the actual experimental processes, including cross-correlated relaxation and unidentified systematic errors. A detailed analysis of ( quantum computer's ) experimental imperfections"

↑ So the error-prone quantum computer's experimental results were worse than the classical computer's numerical results with No quantum computer advantage, contrary to the above fake news.

No quantum computer advantage.

A classical computer numerical simulation easily outperformed the fake quantum computer machine learning (= heavily relying on a classical computer's training data ), contrary to overhyped fake news.

So the fake quantum advantage of this overhyped fake news is caused by the comparison between some bad classical methods (= classical spin reservoir ) and quantum machine learning methods (= QRC ) using just 9 atomic nuclear spins (= Not a quantum computer ) relying on the classical computer's training of weight w with No real quantum computer's advantage over a classical computer.

↑ In this research, the classical computer's numerical simulation (of quantum machine learning methods ) outperformed the (fake) quantum computer of only 9 atomic nuclear spins heavily relying on the classical computer's training of weight w.

↑ This-p.14-Fig.7a-red-line (= classical computer numerical simulation of quantum spin reservoir )'s results R² (= 1 is best ) were better and closer to the perfect prediction of R² than the experimental results of fake quantum computer's machine learning (= quantum reservoir ) in this-p.4-Fig.3a-red line.

Feel free to link to this site.