Researcher auditing AI trustworthiness across generative search evaluation and hardware-security RTL verification
4
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No usable publication-volume data is available in this snapshot, so output is unscored. This axis would measure the count and cadence of indexed publications over time; with only two cataloged arXiv entries from 2026, any productivity estimate would be guesswork.
Impact—
Citation counts, h-index, and downstream-usage signals are absent from the input data, so impact is left unscored. This axis would measure how often and how influentially this researcher's work is cited across the literature.
Focus0/100
The underlying field distribution is maximally concentrated: the bibliometric snapshot assigns all fielded activity to a single field, hardware security and AI for EDA (RTL verification), weighted 1, producing a focus score of 100/100. The cataloged papers reinforce a narrow, audit-style methodological identity, though one strand extends into trustworthy generative search and RAG evaluation.
Collaboration50/100
The snapshot records 5 distinct co-authors, with Kainan Zhou, Chuhong Xu, Zhaoyi Li, Hang Xiao, and Lu Yi recurring; both cataloged 2026 papers place this researcher in a middle authorship position without corresponding-author responsibility. The resulting collaboration score of 36/100 reflects participation in a small, stable team rather than a large or leading network.
Activity77/100
Recent publishing cadence cannot be established: the snapshot contains no year-by-year publication records beyond the two 2026 catalog entries. This axis would measure sustained research activity over recent years, and it is therefore not scored.
Reproducibility66/100
Reproducibility is scored at 55/100, averaged over a single assessed paper (reproPapers: 1). Within the Evolve catalog, the generative-search audit paper carries a high reproducibility-readiness label, while the preregistered RTL verification audit is rated medium after its AI runs yielded no valid analysis — a fail-closed but evidence-thin record.
Research summary
Gangzhen Qian is an early-stage researcher whose cataloged work sits at the intersection of trustworthy generative search evaluation and hardware security in AI-assisted EDA. Two 2026 arXiv papers anchor this profile. The first, 'Claim-Gated Source-Risk Auditing for Generative Search', specifies a four-predicate omission audit for RAG-style answers; Evolve judges it rigorous but self-referential in its conformance tests, and it carries a high reproducibility-readiness rating. The second audits LLM-generated RTL verification plans against authored security-regression proxies; preregistered and fail-closed, its AI runs yielded no valid analysis, so it contributes a benchmark and baseline rather than model-performance evidence. What unifies the two is a methodology-first stance: specify audit predicates in advance, validate the instrument itself, and report null outcomes honestly. Bibliometrically the snapshot is thin: focus scores a maximal 100/100 on a single recorded field (hardware security and AI for EDA, RTL verification), collaboration scores 36/100 on 5 distinct co-authors (Kainan Zhou, Chuhong Xu, Zhaoyi Li, Hang Xiao, Lu Yi), and reproducibility averages 55/100 over one assessed paper. Output, impact, and activity lack underlying data and are left unscored. With no affiliation, country, or ORCID on record, this profile is name-resolved from public bibliographic data.
Collaboration pattern
The snapshot lists 5 distinct co-authors — Kainan Zhou, Chuhong Xu, Zhaoyi Li, Hang Xiao, and Lu Yi — and both 2026 arXiv papers position Gangzhen Qian as a middle author without corresponding responsibility. This suggests a contributor role inside a small, stable team rather than leadership of a large network, consistent with the collaboration score of 36/100. No cross-institution patterns can be inferred because affiliation data is missing.
Research keywords
Highlights
A four-predicate omission audit for generative search
Fail-closed audit of AI-generated RTL verification plans
'Trust, but validate the instrument' as a working method
Across both cataloged papers — one in NLP-flavored search evaluation, one in hardware security — the consistent move is to audit the auditing instrument itself: specify predicates in advance, preregister, and report null outcomes honestly. This fail-closed style underpins the profile's maximal focus score and its early-stage evidence base.
Timeline
- 2026Published 'Claim-Gated Source-Risk Auditing for Generative Search' on arXiv, in trustworthy generative search and RAG evaluation (middle authorship).
- 2026Published 'Trust, but Validate the Instrument: Auditing AI-Generated RTL Verification Plans on Authored Security-Regression Proxies' on arXiv, in hardware security and AI for EDA (middle authorship).
- 2026Collaborated within the five-person co-author pool recorded in the snapshot: Kainan Zhou, Chuhong Xu, Zhaoyi Li, Hang Xiao, and Lu Yi.
Papers in the Evolve catalog
Claim-Gated Source-Risk Auditing for Generative Search
Trust, but Validate the Instrument: Auditing AI-Generated RTL Verification Plans on Authored Security-Regression Proxies
IBBench-Light: A Paired Evaluation of Task-Conditioned Responses to External Directives
Protocol effects on feature-based hardware-Trojan detection across Trust-Hub families
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometrics derive from a thin OpenAlex-based snapshot; affiliation, country, and ORCID are absent, so this profile is name-resolved from public bibliographic data and may merge or split namesakes.
- The reproducibility score (55) averages a single assessed paper, so it should be read as an early-stage indicator rather than a stable estimate.
- Output, impact, and activity are unscored because no publication counts, citation metrics, or year-by-year records were available.
- Both cataloged papers are 2026 arXiv preprints with middle, non-corresponding authorship; peer-reviewed venue status is not recorded.