Hongwei Wang
New York University Abu Dhabi · Mathematics + AI
Hongwei (Alex) Wang
MSc Student in Interdisciplinary Data Science and Artificial Intelligence
Mathematical Foundations of Trustworthy Artificial Intelligence
I work at the intersection of type theory, higher category theory, formalized mathematics, and artificial intelligence, with particular interest in reliable machine reasoning and semantic faithfulness.
- 01 Meaning Mathematical intent
- 02 Formalization Type-theoretic representation
- 03 Verification Machine-checked reasoning
Formal correctness is essential; semantic faithfulness remains the central question.
About
A mathematical approach to reliable reasoning
I am a master's student at New York University Abu Dhabi, with an academic background in pure mathematics. I am particularly interested in how mathematical structures can provide foundations for reliable machine reasoning, and how the meaning of mathematical statements can be preserved when translated between natural language and formal systems.
Proof assistants can mechanically verify formal derivations, but a further question remains: does a formally verified statement faithfully represent the mathematical statement originally intended? This distinction between formal correctness and semantic faithfulness motivates much of my current work.
Research interests
Three connected directions
Mathematical Foundations
Structural languages for objects, proofs, equivalences, and higher-dimensional reasoning.
- Dependent Type Theory
- Homotopy Type Theory
- Higher Category Theory
AI for Formal Mathematics
Examining how mathematical language can become formal code without losing intended meaning.
- Autoformalization
- Semantic Faithfulness
- Machine-Checked Proofs
Trustworthy Artificial Intelligence
Using formal methods to make reasoning systems and AI agents more reliable and verifiable.
- Formal Verification
- Reliable Reasoning
- AI Agent Verification
Current research
Meaning, representation, and machine-verifiable reasoning
I am currently exploring the relationship between mathematical meaning, formal representation, and machine-verifiable reasoning.
- Developing and studying formal representations of mathematical reasoning.
- Investigating semantic faithfulness in AI-assisted mathematical formalization.
- Exploring type-theoretic and homotopy-theoretic approaches to formalized mathematics.
- Applying formal verification techniques to the reliability of AI reasoning systems.
Featured research
Public work and software
Lean 4 · Formalized mathematics
LeanArabic
A feasibility project for representing a published fragment of Arabic pregroup grammar and producing machine-checkable grammatical derivations.
Lean 4 · AI reasoning verification
FalconVerifier
A public verification prototype that separates model-generated interpretation from deterministic formal verification and records machine-checkable assurance artifacts.
Recent news
Academic activity
-
Delivered the talk When Proof No Longer Means Understanding at New York University Abu Dhabi.
-
Began MSc studies in Interdisciplinary Data Science and Artificial Intelligence at New York University Abu Dhabi.
Contact