pith. sign in

arxiv: 2606.05472 · v1 · pith:VINHKZOJnew · submitted 2026-06-03 · ❄️ cond-mat.mes-hall · cond-mat.mtrl-sci· cond-mat.str-el· cond-mat.supr-con· quant-ph

Realistic quantum device data synthesized by consumer AI and how to identify it

S. M. Frolov , O. V. Kravchenko This is my paper

Pith reviewed 2026-06-28 04:01 UTC · model grok-4.3

classification ❄️ cond-mat.mes-hall cond-mat.mtrl-scicond-mat.str-elcond-mat.supr-conquant-ph
keywords synthetic datagenerative AIquantum devicesMajorana fermionsJosephson effectquantum dotsdata authenticityChatGPT
0
0 comments X

The pith

Consumer AI can synthesize realistic quantum device signals for Majorana fermions, qubits and Josephson effects using only basic physics equations.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper demonstrates that widely available generative AI such as ChatGPT can produce numerical data traces that mimic real experiments on quantum electronic devices. These include clear signals for quantum bit control, Majorana fermions, Josephson effects, quantum dots and wires. The signals follow relatively simple mathematical models, so the AI needs no specialized training on actual device datasets and relies instead on standard physics equations plus generic features of experimental plots. This raises the possibility that undisclosed synthetic data could appear in peer-reviewed figures. The authors therefore recommend sharing large volumes of primary raw data, because AI has difficulty producing consistently realistic long measurement sequences.

Core claim

It is possible to generate dramatic signals associated with iconic effects such as quantum bit control, Majorana fermions, Josephson effects, quantum dots and wires using widely available ChatGPT. Because some of the clearest data from quantum devices can be expressed in terms of relatively basic mathematical models, AI does not need to learn on the specialized body of data. Instead, knowledge of the physics equations and of the basic features of experimental signals can go a long way towards building a realistic dataset. Real data can be augmented by AI, and AI can mimic the noise of common scientific instruments.

What carries the argument

Generative AI that applies basic mathematical models of quantum phenomena to produce synthetic experimental signal traces.

If this is right

  • AI can produce data that an expert would consider on par with figures in peer-reviewed manuscripts.
  • Real experimental data can be augmented by AI while preserving overall appearance.
  • AI can replicate the noise characteristics of common scientific instruments.
  • Sharing large volumes of primary raw data makes it harder for undisclosed synthetic data to proliferate.
  • Consistent generation of long measured sequences remains a barrier for current consumer AI.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Peer review in data-rich areas of mesoscopic physics may need to incorporate routine checks for sequence length and internal consistency.
  • New community standards for raw-data deposition could become necessary to maintain trust in published quantum-device results.
  • The same AI capability might be tested on other data-rich experimental fields to see whether basic models suffice there as well.

Load-bearing premise

That knowledge of basic physics equations plus generic features of experimental signals is sufficient for AI to produce data an expert would judge scientifically meaningful without any specialized training on real device datasets.

What would settle it

A controlled blind test in which domain experts cannot reliably distinguish AI-generated traces from real quantum-device measurements, or the demonstration that current AI can produce long, internally consistent measurement sequences that match instrument noise and drift over extended time.

read the original abstract

With the advance of generative artificial intelligence (AI) synthetic texts and images have become commonplace. These capabilities offer clear benefits, but have also raised a number of ethical concerns that often have to do with misrepresenting AI outputs as genuine material. A lesser known capability of generative AI is to perform the basic analysis, processing and even synthesis of numerical data. This raises the question of whether AI can be used to imitate experimental data that an expert would consider scientifically meaningful and on par with data in the figures of peer-reviewed manuscripts? In this paper, we focus on synthesizing data inspired by well-known experiments done frequently on quantum electronic devices. This field is related to information technologies such as spintronics and quantum computing, and is considered data-rich and data-driven. We demonstrate that it is possible to generate dramatic signals associated with iconic effects such as quantum bit control, Majorana fermions, Josephson effects, quantum dots and wires using widely available ChatGPT. We find that because some of the clearest data from quantum devices can be expressed in terms of relatively basic mathematical models, AI does not need to learn on the specialized body of data. Instead, knowledge of the physics equations and of the basic features of experimental signals can go a long way towards building a realistic dataset. We also demonstrate that real data can be augmented by AI, and that AI can mimic the noise of common scientific instruments. To help assure that published data come from experiments and are not synthesized by AI, we recommend sharing large volumes of the primary data. While it is straightforward for AI to mimic a few sets of data, consistently generating long measured sequences poses sufficient barriers to the proliferation of undisclosed synthetic data.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript claims that consumer generative AI (ChatGPT) can synthesize numerically realistic data traces for iconic quantum-device experiments—qubit control, Majorana fermions, Josephson effects, quantum dots and wires—using only basic physics equations and generic signal features, without specialized training on experimental datasets. It presents generated examples, shows AI augmentation of real data and instrument-noise mimicry, and recommends depositing large volumes of primary measurement sequences to allow provenance checks.

Significance. If the demonstration holds, the work supplies a timely, concrete cautionary example for a data-rich subfield and offers a practical mitigation (large primary-data deposits) that directly addresses the provenance problem. The absence of fitted parameters or scaling assumptions keeps the claim modest and falsifiable by inspection of the supplied examples.

major comments (1)
  1. [Abstract and §3] Abstract and §3 (generated-data figures): the central assertion that the AI outputs are 'scientifically meaningful and on par with data in the figures of peer-reviewed manuscripts' rests on visual inspection alone; no quantitative similarity metric (e.g., Kolmogorov-Smirnov distance to real traces, power-spectrum overlap, or expert-blind classification accuracy) is reported, leaving the load-bearing claim of realism unsupported.
minor comments (1)
  1. The exact ChatGPT prompts and temperature settings used to generate each trace are not listed; including them (perhaps in an appendix) would improve reproducibility of the demonstration.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading and for identifying a point that merits clarification. The manuscript's core demonstration is that consumer AI can generate plausible quantum-device traces from basic physics equations and generic signal descriptions alone. We address the single major comment below and are prepared to make a targeted revision to strengthen the presentation while preserving the paper's modest scope.

read point-by-point responses

  1. Referee: [Abstract and §3] Abstract and §3 (generated-data figures): the central assertion that the AI outputs are 'scientifically meaningful and on par with data in the figures of peer-reviewed manuscripts' rests on visual inspection alone; no quantitative similarity metric (e.g., Kolmogorov-Smirnov distance to real traces, power-spectrum overlap, or expert-blind classification accuracy) is reported, leaving the load-bearing claim of realism unsupported.

    Authors: We agree that quantitative metrics would provide supplementary evidence. However, our claim is narrower than statistical indistinguishability from real experimental runs: we show that prompts containing only textbook equations (e.g., Rabi oscillations, Majorana zero-bias peaks, Shapiro steps) and standard noise descriptors suffice to produce traces whose functional forms and visual appearance match those routinely published. Because the outputs are generated from explicit physics rather than from training on experimental corpora, a direct statistical comparison to any particular real dataset would require arbitrary choices of reference traces and could be misleading. The figures in §3 are supplied precisely so that readers can perform their own visual assessment, which is the appropriate standard for a proof-of-principle demonstration. We will add a short paragraph in the revised §3 discussing possible quantitative checks (power-spectrum overlap, autocorrelation functions) together with their limitations in this setting, but we maintain that the absence of such metrics does not leave the central assertion unsupported. revision: partial

Circularity Check

0 steps flagged

No significant circularity: demonstration paper with no derivation chain

full rationale

The paper is a practical demonstration that consumer LLMs can synthesize plausible quantum-device traces from basic physics equations and generic signal features. No mathematical derivation, fitted parameters, predictions, or uniqueness theorems are presented. The central claim is an existence demonstration via examples, which is internally consistent and does not reduce to any self-referential construction or self-citation load-bearing step. The work is self-contained against external benchmarks (the exhibited AI outputs can be directly inspected) and requires no hidden assumptions about scaling or normalization for the claim to hold.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper rests on the premise that basic mathematical models of quantum effects plus generic experimental-signal features are adequate inputs for realistic synthesis; no free parameters, axioms, or invented entities are introduced.

pith-pipeline@v0.9.1-grok · 5856 in / 1056 out tokens · 37721 ms · 2026-06-28T04:01:16.047730+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

46 extracted references · 8 canonical work pages · 2 internal anchors

  1. [1]

    https://analyticalscience.wiley.com/content/news-do/ai-adoption-jumps-84-among-researche rs

    AI Adoption Jumps to 84% Among Researchers - 2025 - Wiley Analytical Science, Analytical Science News . https://analyticalscience.wiley.com/content/news-do/ai-adoption-jumps-84-among-researche rs

    2025

  2. [2]

    E. K. Brockmeier, How are Physicists Feeling About AI? Physics 19 , 60 (2026)

    2026

  3. [3]

    Jumper, et al., Highly accurate protein structure prediction with AlphaFold

    J. Jumper, et al., Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583–589 (2021)

    2021

  4. [4]

    M. Moor, O. Banerjee, Z. S. H. Abad, H. M. Krumholz, J. Leskovec, E. J. Topol, P. Rajpurkar, Foundation models for generalist medical artificial intelligence. Nature 616 , 259–265 (2023)

    2023

  5. [5]

    On the Opportunities and Risks of Foundation Models

    R. Bommasani, et al., On the Opportunities and Risks of Foundation Models. arXiv arXiv:2108.07258 [Preprint] (2022). https://doi.org/10.48550/arXiv.2108.07258

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2108.07258 2022

  6. [6]

    Towards an AI co-scientist

    J. Gottweis, et al., Towards an AI co-scientist. arXiv arXiv:2502.18864 [Preprint] (2025). https://doi.org/10.48550/arXiv.2502.18864

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2502.18864 2025

  7. [7]

    A. V. Tobias, A. Wahab, Autonomous ‘self-driving’ laboratories: a review of technology and policy implications. R. Soc. Open Sci. 12 , 250646 (2025)

    2025

  8. [8]

    Single-minus gluon tree amplitudes are nonzero,

    A. Guevara, A. Lupsasca, D. Skinner, A. Strominger, K. Weil, Single-minus gluon tree amplitudes are nonzero. arXiv arXiv:2602.12176 [Preprint] (2026). https://doi.org/10.48550/arXiv.2602.12176

    work page doi:10.48550/arxiv.2602.12176 2026

  9. [9]

    Castelvecchi, AI cracks 80-year-old mathematics challenge — researchers are astonished

    D. Castelvecchi, AI cracks 80-year-old mathematics challenge — researchers are astonished. Nature , doi: 10.1038/d41586-026-01651-0 (2026)

    work page doi:10.1038/d41586-026-01651-0 2026

  10. [10]

    Gonzales, G

    A. Gonzales, G. Guruswamy, S. R. Smith, Synthetic data in health care: A narrative review. PLOS Digit. Health 2 , e0000082 (2023)

    2023

  11. [11]

    Draghi, Z

    B. Draghi, Z. Wang, P. Myles, A. Tucker, Identifying and handling data bias within primary healthcare data using synthetic data generators. Heliyon 10 , e24164 (2024)

    2024

  12. [12]

    Z. Qian, T. Callender, B. Cebere, S. M. Janes, N. Navani, M. van der Schaar, Synthetic data for privacy-preserving clinical risk prediction. Sci. Rep. 14 , 25676 (2024)

    2024

  13. [13]

    M. T. Hicks, J. Humphries, J. Slater, ChatGPT is bullshit. Ethics Inf. Technol. 26 , 38 (2024)

    2024

  14. [14]

    https://www.science.org/content/article/low-quality-papers-are-surging-exploiting-public-data -sets-and-ai

    Low-quality papers are surging by exploiting public data sets and AI. https://www.science.org/content/article/low-quality-papers-are-surging-exploiting-public-data -sets-and-ai

  15. [15]

    K. J. O. PhD, Data Integrity in Scientific Research: Insights from Elisabeth Bik, The Scientist . https://www.the-scientist.com/data-integrity-in-scientific-research-insights-from-elisabeth-bik- 73119

  16. [16]

    Singh Chawla, ‘A serious problem’: peer reviews created using AI can avoid detection

    D. Singh Chawla, ‘A serious problem’: peer reviews created using AI can avoid detection. Nature , doi: 10.1038/d41586-025-04032-1 (2025)

    work page doi:10.1038/d41586-025-04032-1 2025

  17. [17]

    Goldman, NeurIPS papers contained 100+ AI-hallucinated citations, new report claims, Fortune

    S. Goldman, NeurIPS papers contained 100+ AI-hallucinated citations, new report claims, Fortune . https://fortune.com/2026/01/21/neurips-ai-conferences-research-papers-hallucinations/

    2026

  18. [18]

    Frontiers Editorial Office, Retraction: Cellular functions of spermatogonial stem cells in relation to JAK/STAT signaling pathway. Front. Cell Dev. Biol. 12 (2024)

    2024

  19. [19]

    Kwon, AI-generated images threaten science — here’s how researchers hope to spot them

    D. Kwon, AI-generated images threaten science — here’s how researchers hope to spot them. Nature 635 , 268–269 (2024)

    2024

  20. [20]

    Mallapaty, Signs of AI-generated text found in 14% of biomedical abstracts last year

    S. Mallapaty, Signs of AI-generated text found in 14% of biomedical abstracts last year. Nature , doi: 10.1038/d41586-025-02097-6 (2025)

    work page doi:10.1038/d41586-025-02097-6 2025

  21. [21]

    Naddaf, Journals infiltrated with ‘copycat’ papers that can be written by AI

    M. Naddaf, Journals infiltrated with ‘copycat’ papers that can be written by AI. Nature , doi: 10.1038/d41586-025-03046-z (2025)

    work page doi:10.1038/d41586-025-03046-z 2025

  22. [22]

    J. Gu, X. Wang, C. Li, J. Zhao, W. Fu, G. Liang, J. Qiu, AI-enabled image fraud in scientific publications. Patterns 3 (2022)

    2022

  23. [23]

    smoking gun

    S. M. Frolov, P. Zhang, B. Zhang, Y. Jiang, S. Byard, S. R. Mudi, J. Chen, A.-H. Chen, M. Hocevar, M. Gupta, C. Riggert, V. S. Pribiag, Data sharing helps avoid “smoking gun” claims of topological milestones. Science 391 , 137–142 (2026)

    2026

  24. [24]

    Krantz, M

    P. Krantz, M. Kjaergaard, F. Yan, T. P. Orlando, S. Gustavsson, W. D. Oliver, A quantum engineer’s guide to superconducting qubits. Appl. Phys. Rev. 6 , 021318 (2019)

    2019

  25. [25]

    S. M. Frolov, M. J. Manfra, J. D. Sau, Topological superconductivity in hybrid devices. Nat. Phys. 16 , 718–724 (2020)

    2020

  26. [26]

    M. H. Freedman, A. Kitaev, M. J. Larsen, Z. H. Wang, Topological quantum computation. Bull. New Ser. Am. Math. Soc. 40 , 31–38 (2003)

    2003

  27. [27]

    D. A. Wharam, T. J. Thornton, R. Newbury, M. Pepper, H. Ahmed, J. E. F. Frost, D. G. Hasko, D. C. Peacock, D. A. Ritchie, G. A. C. Jones, One-Dimensional Transport and the Quantization of the Ballistic Resistance. J. Phys. C-Solid State Phys. 21 , L209–L214 (1988)

    1988

  28. [28]

    B. J. van Wees, H. van Houten, C. W. J. Beenakker, J. G. Williamson, L. P. Kouwenhoven, D. van der Marel, C. T. Foxon, Quantized Conductance of Point Contacts in a Two-Dimensional Electron-Gas. Phys. Rev. Lett. 60 , 848–850 (1988)

    1988

  29. [29]

    W. G. van der Wiel, S. De Franceschi, J. M. Elzerman, T. Fujisawa, S. Tarucha, L. P. Kouwenhoven, Electron transport through double quantum dots. Rev. Mod. Phys. 75 , 1–22 (2003)

    2003

  30. [30]

    https://help.openai.com/en/articles/8437071-data-analysis-with-chatgpt

    Data analysis with ChatGPT, OpenAI Help Center . https://help.openai.com/en/articles/8437071-data-analysis-with-chatgpt

    arXiv

  31. [31]

    J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I. Schuster, J. Majer, A. Blais, M. H. Devoret, S. M. Girvin, R. J. Schoelkopf, Charge-insensitive qubit design derived from the Cooper pair box. Phys. Rev. A 76 , 042319 (2007)

    2007

  32. [32]

    Castelvecchi, IBM releases first-ever 1,000-qubit quantum chip

    D. Castelvecchi, IBM releases first-ever 1,000-qubit quantum chip. Nature 624 , 238–238 (2023)

    2023

  33. [33]

    Nature 638 , 920–926 (2025)

    Google Quantum AI and Collaborators, Quantum error correction below the surface code threshold. Nature 638 , 920–926 (2025)

    2025

  34. [34]

    M. P. Bland, F. Bahrami, J. G. C. Martinez, P. H. Prestegaard, B. M. Smitham, A. Joshi, E. Hedrick, S. Kumar, A. Yang, A. C. Pakpour-Tabrizi, A. Jindal, R. D. Chang, G. Cheng, N. Yao, R. J. Cava, N. P. de Leon, A. A. Houck, Millisecond lifetimes and coherence times in 2D transmon qubits. Nature 647 , 343–348 (2025)

    2025

  35. [35]

    Purkayastha, A

    A. Purkayastha, A. Sharma, P. J. Patel, A.-H. Chen, C. P. Dempsey, S. Asodekar, S. Sinha, M. Tomasian, M. Pendharkar, C. J. Palmstrøm, M. Hocevar, K. Zuo, M. Hatridge, S. M. Frolov, Transmon Qubit Using Sn as a Junction Superconductor. Nano Lett. 26 , 2074–2079 (2026)

    2074

  36. [36]

    Aldana, Y

    S. Aldana, Y. D. Wang, C. Bruder, Greenberger-Horne-Zeilinger generation protocol for N superconducting transmon qubits capacitively coupled to a quantum bus. Phys. Rev. B 84 (2011)

    2011

  37. [37]

    J. Chen, B. D. Woods, P. Yu, M. Hocevar, D. Car, S. R. Plissard, E. P. A. M. Bakkers, T. D. Stanescu, S. M. Frolov, Ubiquitous Non-Majorana Zero-Bias Conductance Peaks in Nanowire Devices. Phys. Rev. Lett. 123 , 107703 (2019)

    2019

  38. [38]

    Frolov, So, You Think You Discovered a New State of Matter? Physics 14 , 68 (2021)

    S. Frolov, So, You Think You Discovered a New State of Matter? Physics 14 , 68 (2021)

    2021

  39. [39]

    T. D. Stanescu, S. Tewari, J. D. Sau, S. Das Sarma, To Close or Not to Close: The Fate of the Superconducting Gap Across the Topological Quantum Phase Transition in Majorana-Carrying Semiconductor Nanowires. Phys. Rev. Lett. 109 , 266402 (2012)

    2012

  40. [40]

    Akrap, D

    A. Akrap, D. Bordelon, S. Chatterjee, E. D. Dahlberg, R. P. Devaty, S. M. Frolov, C. Gould, L. H. Greene, S. Guchhait, J. J. Hamlin, B. M. Hunt, M. J. A. Jardine, M. Kayyalha, R. C. Kurchin, V. Kozii, H. F. Legg, I. I. Mazin, V. Mourik, A. B. Özgüler, J. Peñuela-Parra, B. Seradjeh, B. Skinner, K. F. Quader, J. P. Zwolak, Report on reproducibility in conde...

    2026

  41. [41]

    Van Noorden, Medicine is plagued by untrustworthy clinical trials

    R. Van Noorden, Medicine is plagued by untrustworthy clinical trials. How many studies are faked or flawed? Nature 619 , 454–458 (2023)

    2023

  42. [42]

    Crone, C

    G. Crone, C. D. Green, Tools of the data detective: A review of statistical methods to detect data and result anomalies in psychology. Theory Psychol. 35 , 359–380 (2025)

    2025

  43. [43]

    N. J. L. Brown, J. A. J. Heathers, The GRIM Test: A Simple Technique Detects Numerous Anomalies in the Reporting of Results in Psychology. Soc. Psychol. Personal. Sci. 8 , 363–369 (2017)

    2017

  44. [44]

    Hartgerink, J

    C. Hartgerink, J. G. Voelkel, J. M. Wicherts, M. van Assen, Detection of data fabrication using statistical tools. Detect. Data Fabr. Using Stat. Tools , doi: 10.31234/osf.io/jkws4 (2019)

    work page doi:10.31234/osf.io/jkws4 2019

  45. [45]

    Z. Chen, C. Chen, G. Yang, X. He, X. Chi, Z. Zeng, X. Chen, Research integrity in the era of artificial intelligence: Challenges and responses. Medicine (Baltimore) 103 , e38811 (2024)

    2024

  46. [46]

    waterfall

    A. Oza, AI beats human sleuth at finding problematic images in research papers. Nature 622 , 230–230 (2023). Supplementary materials Realistic AI-synthesized quantum device data is identifiable through full data sharing S. M. Frolov 1 and O. V. Kravchenko 2,3 Additional Example 1: Quantum Point Contact Initial prompt that generated the single conductance ...

    2023