• Computer Science > Artificial Intelligence [Submitted on 24 Feb 2026] Title:Physics-based phenomenological characterization of cross-modal bias in multimodal models View PDF HTML (experimental)Abstract:The term ‘algorithmic fairness’ is used to evaluate whether AI models operate fairly in both comparative (where fairness is understood as formal equality, such as “treat like cases as like”) and non-comparative (where unfairness arises from the model’s inaccuracy, arbitrariness, or inscrutability) contexts. • Recent advances in multimodal large language models (MLLMs) are breaking new ground in multimodal understanding, reasoning, and generation; however, we argue that inconspicuous distortions arising from complex multimodal interaction dynamics can lead to systematic bias. • The purpose of this position paper is twofold: first, it is intended to acquaint AI researchers with phenomenological explainable approaches that rely on the physical entities that the machine experiences during training/inference, as opposed to the traditional cognitivist symbolic account or metaphysical approaches; second, it is to state that this phenomenological doctrine will be practically useful for tackling algorithmic fairness issues in MLLMs. • We develop a surrogate physics-based model that describes transformer dynamics (i.e., semantic network structure and self-/cross-attention) to analyze the dynamics of cross-modal bias in MLLM, which are not fully captured by conventional embedding- or representa

Article Summaries:

  • Physics‑based Phenomenological Approach to Cross‑Modal Bias in Multimodal Models

A recent position paper proposes a new framework for assessing algorithmic fairness in multimodal large language models (MLLMs). Rather than relying on symbolic or purely statistical analyses, the authors introduce a physics‑based surrogate model that captures transformer dynamics-semantic network structure, self‑ and cross‑attention-to study how multimodal inputs can create systematic bias. Experiments on Qwen2.5‑Omni, Gemma 3n, and Lorenz‑chaotic time‑series prediction show that multimodal signals can reinforce modality dominance, producing structured error‑attractor patterns under label perturbation. The paper argues that this phenomenological perspective can guide practical fairness interventions in MLLMs.

Sources: