Sony researcher auditing AI instruments: LLM-generated RTL verification plans and claim-gated generative search
5
papers here
5
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—
Output cannot be scored yet: the OpenAlex snapshot reports no populated productivity statistics (h-index 0, i10-index 0). This axis would measure cumulative works per year and venue spread once publication indexing for this author is complete.
Impact—
No citation data is available to score impact. This axis would measure citations, field-normalized influence, and uptake of the two 2026 audit papers once citation indexing accumulates around this name-resolved profile.
Focus0/100
Scores 100 for concentration: the OpenAlex field distribution lists a single field — hardware security and AI for EDA (RTL verification) — and the leading topics, Advanced Malware Detection Techniques and Security and Verification in Computing, hold 2 works each. Both cataloged 2026 papers share one move, auditing AI systems as instruments, so the measured focus looks genuinely narrow rather than an artifact of sparse indexing.
Collaboration55/100
Scores 36, indicating a compact network: OpenAlex records 5 distinct coauthors, and the recurring group comprises Kainan Zhou, Gangzhen Qian, Zhaoyi Li, Hang Xiao, and Lu Yi. On both cataloged 2026 papers Chuhong Xu appears in a middle author position without corresponding-author status, consistent with contributing specialist roles inside small teams rather than leading large consortia.
Activity86/100
Scores 57 on current output rhythm: the snapshot records 6 works in the recent three-year window (recent3yWorks: 6), and the counts-by-year series places all 6 in 2026. Both Evolve-reviewed papers appeared as 2026 arXiv preprints, so measured activity is recent and concentrated in a single year, with no intermediate-year signal available to confirm a sustained pace.
Reproducibility66/100
Scores 55, computed from 1 assessed paper (reproMean 55, reproPapers 1). The generative-search audit reaches high reproducibility readiness through a fully specified four-predicate conformance test, while the preregistered RTL verification audit is rated medium after its AI runs yielded no valid analysis yet still shipped a fail-closed benchmark and baseline. The pattern favors explicit, checkable procedures over unverifiable claims.
Research summary
Chuhong Xu is a researcher at Sony Corporation (United States) whose cataloged work converges on a single methodological stance: treat AI systems as instruments that must themselves be audited before their outputs are trusted. Name-resolved from OpenAlex bibliographic data (ORCID 0009-0007-3447-2558), the profile rests on two 2026 arXiv preprints reviewed by Evolve. The first, Claim-Gated Source-Risk Auditing for Generative Search, specifies a four-predicate omission audit for retrieval-augmented generation, pairing rigorous conformance tests with an explicit deferral of semantic validity to future work; Evolve rates its reproducibility readiness high. The second, Trust, but Validate the Instrument, preregisters an audit of LLM-generated RTL verification plans against authored security-regression proxies; its AI runs yielded no valid analysis, yet the fail-closed benchmark and baseline it released are judged valuable, with medium reproducibility readiness. OpenAlex topics place adjacent work in advanced malware detection and security and verification in computing (2 works each), with single works in parallel computing, information retrieval, and video coding. The snapshot records 6 works in the recent three-year window, 5 distinct coauthors, and no populated citation indices, so output and impact remain unscored. The consistent signature — preregistration, explicit predicates, fail-closed reporting — marks an early-stage but methodologically disciplined profile.
Collaboration pattern
OpenAlex records 5 distinct coauthors, and the same five names recur: Kainan Zhou, Gangzhen Qian, Zhaoyi Li, Hang Xiao, and Lu Yi. Both cataloged 2026 papers place Chuhong Xu in a middle author position without corresponding status, suggesting a contributor role in small, focused teams; the collaboration dimension scores 36, consistent with a compact, project-driven network rather than a large consortium.
Research keywords
Highlights
Four-predicate omission audit for generative search
Preregistered audit of LLM-generated RTL verification plans
Fail-closed evaluation as a recurring signature
Across both 2026 catalog entries the pattern repeats: specify checks before running AI systems, then report null results openly. The pair earns one high and one medium reproducibility readiness rating, and the reproducibility dimension scores 55 from a single assessed paper — an early-stage but disciplined evaluation posture.
Timeline
- 2026Claim-Gated Source-Risk Auditing for Generative Search appears on arXiv; Evolve rates its reproducibility readiness high.
- 2026Trust, but Validate the Instrument preregisters an audit of LLM-generated RTL verification plans; the AI runs yield no valid analysis, and a fail-closed benchmark plus baseline is released.
- 2026The OpenAlex snapshot records 6 works in the recent three-year window (recent3yWorks: 6), with the counts-by-year series placing all 6 in 2026.
- 2026The topic profile consolidates around security: Advanced Malware Detection Techniques and Security and Verification in Computing each count 2 works.
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
Query-Aware Source-Risk Triage for Retrieval-Augmented Generation
GGUF-Metadata Prediction of Single-Sequence llama.cpp Throughput Across Three Systems
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 come from an OpenAlex snapshot only (identity: openalex); its citation indices show h-index 0 and i10-index 0, so the output and impact axes are left unscored rather than estimated.
- The counts-by-year series is sparse (1 work in 2006, 6 in 2026), so long-term productivity trends cannot be characterized from this data.
- The qualitative assessment rests on 2 Evolve-reviewed 2026 arXiv preprints with middle authorship, and the reproducibility score averages a single assessed paper.
- Chuhong Xu is a relatively common name; this profile is name-resolved from public bibliographic data (ORCID 0009-0007-3447-2558) and may not capture every work by the researcher.