Ahsan Habib ShihabResearch & practice

01RESEARCH & PUBLICATIONS

Research on
AI evaluation & reliability.

Language model evaluation, multilingual NLP, and AI reliability. Published work, accepted papers, presentations, and work under review—with their status kept explicit.

Papers & presentations

Updated 3 November 2026

2026Published

Multilingual evaluation

Same Problem, Different Answer: Accuracy Hides Cross-Lingual Inconsistency in Multilingual Inference 

A. H. Shihab, A. Ospan, A. Mussa, and M. Mansurova

Proceedings of the 3rd Workshop on Language Understanding in the Human-Machine Era (LUHME 2026) · Association for Computational Linguistics · Budapest, Hungary · pp. 246–255

An item-level assessment of cross-lingual consistency in two zero-shot multilingual natural language inference models on the parallel XNLI benchmark, alongside per-language accuracy.

Read paper Code
Cite Same Problem, Different Answer: Accuracy Hides Cross-Lingual Inconsistency in Multilingual Inference

A. H. Shihab, A. Ospan, A. Mussa, and M. Mansurova. Same Problem, Different Answer: Accuracy Hides Cross-Lingual Inconsistency in Multilingual Inference. Proceedings of the 3rd Workshop on Language Understanding in the Human-Machine Era (LUHME 2026) · Association for Computational Linguistics · Budapest, Hungary · pp. 246–255 (2026). https://aclanthology.org/2026.luhme-1.19/

2026Published

Evaluation methodology

Circularity-Controlled Evaluation: An Independent Assessment Methodology for AI Governance and Oversight Mechanisms 

A. H. Shihab, A. Mussa, A. Ospan, and A. Delavar

2026 IEEE 24th International Symposium on Intelligent Systems and Informatics (SISY) · Pula, Croatia · pp. 397–404

A conceptual methodology separating governance mechanisms from their assessment through evaluation independence, outcome-based assessment, blind controls, and counterfactual assessment.

Presented. The paper illustrates the methodology through a qualitative case study; it does not validate the governance framework itself.

Read paper
Cite Circularity-Controlled Evaluation: An Independent Assessment Methodology for AI Governance and Oversight Mechanisms

A. H. Shihab, A. Mussa, A. Ospan, and A. Delavar. Circularity-Controlled Evaluation: An Independent Assessment Methodology for AI Governance and Oversight Mechanisms. 2026 IEEE 24th International Symposium on Intelligent Systems and Informatics (SISY) · Pula, Croatia · pp. 397–404 (2026). https://doi.org/10.1109/SISY71351.2026.11738459

2026Published

Ensemble evaluation

How Many Advisors Are Enough? Predicting Useful Majority-Vote Ensemble Size under Correlated Decision Errors 

A. H. Shihab, A. Mussa, and M. Mansurova

2026 IEEE 13th International Conference on Intelligent Systems (IS) · Varna, Bulgaria · pp. 1–7

A Beta-binomial model of exchangeable binary advisors for predicting useful majority-vote ensemble size from advisor accuracy and pairwise error correlation.

Presented on 4 September 2026.

Keywords: Ensemble methods; majority voting; correlated errors; Beta-binomial distribution; effective sample size; decision aggregation.

Read paper Code
Cite How Many Advisors Are Enough? Predicting Useful Majority-Vote Ensemble Size under Correlated Decision Errors

A. H. Shihab, A. Mussa, and M. Mansurova. How Many Advisors Are Enough? Predicting Useful Majority-Vote Ensemble Size under Correlated Decision Errors. 2026 IEEE 13th International Conference on Intelligent Systems (IS) · Varna, Bulgaria · pp. 1–7 (2026). https://doi.org/10.1109/IS71416.2026.11727194

2026Published

Multilingual evaluation

When Does Fine-Tuning Matter for a Low-Resource Language? Evaluating Open-Weight LLMs on Kazakh Benchmarks 

Aman Mussa, Zhanseit Tuimebayev, Madina Mansurova, Ahsan Habib Shihab

Information · 17(9), 921

An examination of fine-tuning and open-weight language model evaluation in Kazakh.

Read paper
Cite When Does Fine-Tuning Matter for a Low-Resource Language? Evaluating Open-Weight LLMs on Kazakh Benchmarks

Aman Mussa, Zhanseit Tuimebayev, Madina Mansurova, Ahsan Habib Shihab. When Does Fine-Tuning Matter for a Low-Resource Language? Evaluating Open-Weight LLMs on Kazakh Benchmarks. Information · 17(9), 921 (2026). https://doi.org/10.3390/info17090921

2026Published

Agentic AI governance

When Identical Conditions Yield Different Authority: Measuring Path-Dependence in Trust-to-Authorization Policies 

Ahsan Habib Shihab, Aman Mussa, Madina Mansurova

MIWAI 2026 · Springer, LNCS · pp. 259–272

Examining whether the route to a decision changes the authority assigned by trust-to-authorization policies.

Read paper
Cite When Identical Conditions Yield Different Authority: Measuring Path-Dependence in Trust-to-Authorization Policies

Ahsan Habib Shihab, Aman Mussa, Madina Mansurova. When Identical Conditions Yield Different Authority: Measuring Path-Dependence in Trust-to-Authorization Policies. MIWAI 2026 · Springer, LNCS · pp. 259–272 (2026). https://doi.org/10.1007/978-981-92-6510-7_20

2026Published

Privacy & climate resilience

Privacy-Preserving Federated Learning for Post-Flood Security and Resource Planning in Bangladesh: A Review and Conceptual Framework 

R. A. Druba, A. H. Shihab, A. Mussa, M. Mansurova

IEEE QPAIN 2026

A review and conceptual framework connecting distributed learning, data privacy, and post-flood planning.

Read paper
Cite Privacy-Preserving Federated Learning for Post-Flood Security and Resource Planning in Bangladesh: A Review and Conceptual Framework

R. A. Druba, A. H. Shihab, A. Mussa, M. Mansurova. Privacy-Preserving Federated Learning for Post-Flood Security and Resource Planning in Bangladesh: A Review and Conceptual Framework. IEEE QPAIN 2026 (2026). https://doi.org/10.1109/QPAIN69676.2026.11546266

2026Presented

Benchmark methodology

How Many Prompts Certify a Ranking? A Generalizability Theory Protocol for Compute-Optimal, Dependable LLM Benchmark Evaluation

ICCBD 2026

Presented. Proceedings and DOI are not yet confirmed.

Code
Cite How Many Prompts Certify a Ranking? A Generalizability Theory Protocol for Compute-Optimal, Dependable LLM Benchmark Evaluation

Author metadata is not confirmed in this record. This is a partial reference.

How Many Prompts Certify a Ranking? A Generalizability Theory Protocol for Compute-Optimal, Dependable LLM Benchmark Evaluation. ICCBD 2026 (2026).

2026Accepted · camera-ready

Agentic AI governance

Specification–Behavior Gaps in Trust-Based Authorization: A Gated-Quantity Inertness Diagnostic and Its Limits

CSAI 2026

Accepted/camera-ready. Presentation mode and DOI are not yet confirmed.

Cite Specification–Behavior Gaps in Trust-Based Authorization: A Gated-Quantity Inertness Diagnostic and Its Limits

Author metadata is not confirmed in this record. This is a partial reference.

Specification–Behavior Gaps in Trust-Based Authorization: A Gated-Quantity Inertness Diagnostic and Its Limits. CSAI 2026 (2026).

2026Presented

Sensitive-domain evaluation

Risk-Aware Evaluation of Domain-Adaptive Language Models in Sensitive Public-Sector Applications

AICT 2026 · 14–16 October

Presented. Publication status and DOI are not yet confirmed.

Cite Risk-Aware Evaluation of Domain-Adaptive Language Models in Sensitive Public-Sector Applications

Author metadata is not confirmed in this record. This is a partial reference.

Risk-Aware Evaluation of Domain-Adaptive Language Models in Sensitive Public-Sector Applications. AICT 2026 · 14–16 October (2026).

2026Published

AI reliability

Towards Trustworthy AI in Energy Systems: A Risk-Aware Evaluation Framework for Safe Deployment in Critical Infrastructure 

A. H. Shihab, A. Mussa, and M. Mansurova

Engineering Proceedings, 160(1), 50

Presented in person at NURER 2026.

Read paper
Cite Towards Trustworthy AI in Energy Systems: A Risk-Aware Evaluation Framework for Safe Deployment in Critical Infrastructure

A. H. Shihab, A. Mussa, and M. Mansurova. Towards Trustworthy AI in Energy Systems: A Risk-Aware Evaluation Framework for Safe Deployment in Critical Infrastructure. Engineering Proceedings, 160(1), 50 (2026). https://doi.org/10.3390/engproc2026160050

Accepted

Agentic AI governance

The Verdict Is in the Protocol: A Circularity-Controlled Evaluation Methodology for Runtime Governance of Agentic AI

A. H. Shihab, M. Mansurova, A. Kozierkiewicz, A. Mussa, and T. Sarsembayeva

IEEE Access

Accepted at IEEE Access. Publication details and DOI are not yet confirmed.

Cite The Verdict Is in the Protocol: A Circularity-Controlled Evaluation Methodology for Runtime Governance of Agentic AI

A. H. Shihab, M. Mansurova, A. Kozierkiewicz, A. Mussa, and T. Sarsembayeva. The Verdict Is in the Protocol: A Circularity-Controlled Evaluation Methodology for Runtime Governance of Agentic AI. IEEE Access.

A presentation or acceptance does not imply proceedings publication. Related research questions may appear across multiple works.

02RESEARCH DIRECTIONS

Questions connected to the work.

These directions connect papers, presentations and project dossiers. Each record keeps its own status and evidence limits.

03RESEARCH AS A SHARED PRACTICE

Reading. Reviewing.
Sharing methods.

Academic service

Peer reviewer for IEEE AICT 2026 and IEEE SISY 2026. Five papers reviewed for AICT.

Teaching & exchange

Hands-on multi-agent AI workflow sessions at KazNU’s AI Summer School, using Pronto Sage with visiting students from China.

↗CONTACT & COLLABORATION

Have a question
in common?

I welcome conversations about research, graduate study, and collaborations in AI evaluation, multilingual NLP, and reliable AI systems.