Lei Gao / Research
Research & public outputs

Selected work

A curated record of publications, conference work and public pre-standardization / governance contributions. Entries are labelled so self-maintained research pages are not confused with adopted institutional positions.

Research taxonomy

Research areas and operational vocabulary

Core programme
Operational Human OversightAgentic AI GovernanceAI Governance & Accountability
Operational mechanisms
Human authorityIntervention triggersEscalationQualified reviewOverride & suspensionRevalidation & resumptionEvidence & auditability
Application settings
Public-sector AISmart Sustainable CitiesInstitutional disruption & crisis
Methods & theory
System DynamicsAlgorithmic Friction
Publication

Finance Research Letters

CEO digital background and corporate digitalization: the role of regional policy. Finance Research Letters, Volume 88 (2026), Article 109135.

Publisher record: DOI / ScienceDirect ↗

Conference

International System Dynamics Conference 2026

Algorithmic Friction and Yin-Yang Cybernetics: Reconstructing the Foundational Logic of System Dynamics in the Age of AI.

The official ISDC schedule lists Lei Gao with this presentation. Conference schedule ↗

FG-TIDA

Theme #16 — Operational Human Oversight Integration for Agentic AI Trust Systems

Public working proposal focused on intervention triggers, escalation, qualified human review, human override, suspension, evidence requirements and return-to-operation conditions in agentic AI trust systems.

The public FG-TIDA themes repository identifies Lei Gao as proposer. This is an open working theme, not an adopted ITU Recommendation. Theme #16 ↗

As proposer, Lei Gao has contributed to working-structure synthesis, candidate requirements drafting and coordination of technical discussion. This describes contributions within an open working process, not a formal Working Group leadership appointment.

30 October 2026 working update

Theme #16 — current working position

Public working proposalProposer: Lei GaoStatus checked: 30 October 2026
  • Working baseline: Consolidated Working Structure v1; Requirements Matrix v0.2 and the published UC-6 calibration remain unchanged.
  • UC-6 Stage 0: a bounded, reproducible calibration of selected mapping semantics only. It does not establish meaningful or authenticated human review, operational oversight capacity, live enforcement or successful control transfer.
  • UC-4: deferred-action mapping remains a candidate and depends on an agreed versioned source case and expected observations.
  • Candidate D1 Requirements v0.1: shared for technical review; it does not replace Matrix v0.2 and is not an adopted requirement set.
  • Operator-facing assessment: the current term is “operator-facing experimental assessment package”, rather than “validation package”. Its prerequisites — the versioned source case, expected observations, protocol scope, independence arrangements, fallback conditions and closure criteria — remain to be agreed. It is not experiment-ready and is not evidence of validated oversight effectiveness.

The discussion keeps evidence availability and presentation, human notice and understanding, technical execution and acknowledgement, and independently observed effects distinct. A fallback should specify its trigger, responsible authority and expected effect.

As proposer, Lei Gao has contributed to consolidating the working structure, preparing Candidate D1 requirements for technical review, and helping coordinate discussion of use-case prerequisites and the separately scoped operator-facing assessment. This is a contribution within an open working process, not a formally appointed Working Group leadership role.

These are preparatory working materials, not an adopted ITU-T specification, Working Group charter, formal deliverable or institutional decision.

Theme #16 public thread ↗ · Candidate D1 requirements note ↗ · Operator-facing assessment discussion ↗

FG-AI4SSC

FGAI4SSC-I-015 — Human Oversight in AI-enabled Smart Sustainable Cities

Formal input document to ITU-T FG-AI4SSC, focused on operational human oversight in AI-enabled urban systems. The contribution has subsequently been assigned to WG4 for follow-up within the Focus Group process.

WG4 assignment is a process status and does not imply adoption, endorsement, an ITU Recommendation or standard. The public overview below is self-maintained. Public research overview ↗ · Official ITU cities / FG-AI4SSC context ↗

U4SSC · WG8

U4SSC-DT-WG8-DOC039 — Operational Human Oversight

Working-group contribution within U4SSC Digital Transformation WG8, extending operational human-oversight mechanisms into people-centred digital and AI-enabled city services.

This site describes DOC039 as a working-group contribution / working document. It does not present it as an adopted U4SSC standard, Recommendation or endorsement. U4SSC official initiative ↗

Selected research questions

Questions guiding the current research programme

These are developing research questions, not reported empirical findings.

Persuasive but incorrect AI adviceWhen AI advice is persuasive but wrong, under what conditions can actionable human oversight reduce over-reliance and improve decision quality?
Unavailable or ineffective supervisionWhat should an agentic AI system do when legitimate human supervision becomes unavailable, unreliable or too slow to intervene?
Revalidation before resumptionHow should delegated authority, evidence and system state be revalidated before AI-supported operations resume?
Restoring legitimate controlHow can legitimate human decision authority and accountability be restored when institutions, personnel or evidence channels are disrupted?
Research collaboration

Contributions to collaborative research teams

My strongest contribution is at the interface between governance concepts and operational research design:

Research framing and synthesisTranslate broad governance concerns into bounded research questions, scopes and structured working documents.
Governance-to-requirements mappingOrganise roles, decision authority, intervention triggers, escalation, review, fallback and evidence requirements into checkable structures.
Scenario and assessment designDevelop bounded use cases, observable checkpoints, evidence needs and evaluation boundaries without overstating validation.
Research coordinationOrganise cross-disciplinary input, versioned drafts, contributor attribution, technical review and written deliverables.

These are research-design and coordination contributions. They do not imply responsibility for production-system implementation, independent certification or underlying model development.

Emerging research agenda

Human Oversight under Institutional Disruption and Crisis

This line extends the core Operational Human Oversight programme into high-disruption settings, including prolonged emergencies and the recovery of essential public services. The central question remains how legitimate human control can be preserved, constrained and restored when ordinary oversight conditions fail.

In developmentInstitutional resilienceAuthority recoveryPublic-service continuity

A developing research agenda, not a completed project, validated framework or published empirical case.

Research themes

Current research vocabulary

  • Operational Human Oversight
    Intervention triggers, escalation, qualified review, override, suspension, resumption and evidence.
  • Agentic AI Governance
    Delegation, bounded autonomy, trust lifecycles and accountability.
  • Smart Sustainable Cities
    Human authority, urban AI systems and auditability.
  • System Dynamics / Algorithmic Friction
    How governance points preserve judgment and accountability in automated systems.
Future public additions may include selected talks, conference records and de-identified applied outputs. Confidential client materials and raw enterprise data will not be published.