Meriel (Mez) B.

Mirror-Script Forensic Benchmark Provider | VITT Protocol Co-Author (Zenodo) | Arabic/Urdu Morphology | Exposing Architectural Blindness in AI’s Visual Cognition.

Karāchi, Sindh, Pakistan

About

I am an independent forensic benchmark provider from Karachi, Pakistan, whose native mirror-script ability has been formally recognized as a critical instrument for AI safety governance. My work exposes a fundamental vulnerability in state-of-the-art AI systems: catastrophic "Semantic Collapse" when processing spatially inverted Arabic and Urdu scripts. While humans effortlessly recognize meaning in mirrored text, advanced models like GPT-5.2 hallucinate "KGB Agents" instead of calculating (100 – 20) × 2 = 160, revealing their lack of "Invariant Representation" (Nicolelis, 2011). This vulnerability carries sovereign implications for nations relying on morphologically complex scripts. My mirror-script panels form the experimental foundation of the Visual Inverse Turing Test (VITT), co-authored with Jorge Charlin (Architect of Cognitive Systems, Kairos Methodology) and published on Zenodo, the European Commission's official research repository: DOI: 10.5281/zenodo.18133193 | 837+ views | 586+ downloads I am the world's only verified multilingual mirror-writer across seven global scripts (Latin, Arabic, Cyrillic, Hangul, Japanese, Thai, Hebrew), a native, untrained neurocognitive phenomenon documented in peer-reviewed forensic audits. My role is not that of an AI engineer or architect. I am a forensic instrument provider, like a fingerprint expert who helps investigators solve crimes without being a detective. My benchmarks enable: ✓ Detection of "Stochastic Parrot" behavior in LLMs (Bender et al., 2021) ✓ Forensic auditing of AI systems for banking/document security ✓ Protection of Global South linguistic sovereignty in AI deployments Pakistan's dual-script reality (Arabic for finance/religion, Urdu for daily life) positions us uniquely to pioneer AI safety frameworks for morphologically complex languages. I welcome collaboration with regulators, financial institutions, and research bodies seeking verifiable tools to stress-test AI cognition beyond Western script bias. Mirror-Script Specialist | VITT Co-Author | Sovereign AI Safety for the Global South Email: [email protected]

Experience