AI ethics on Saturday, 7 March 2026

Biases in large language models (LLMs) often manifest as systematic distortions in associations between demographic attributes and professional or social roles, reinforcing harmful stereotypes across gender, ethnicity, and geography. This position paper advocates for addressing demographic and gender biases in LLMs through a dual-pronged methodology, integrating category-theoretic transformations and retrieval-augmented generation (RAG). Category theory provides a rigorous, structure-preserving

Concept Bottleneck Models (CBMs) aim for ante-hoc interpretability by learning a bottleneck layer that predicts interpretable concepts before the decision. State-of-the-art approaches typically select which concepts to learn via human specification, open knowledge graphs, prompting an LLM, or using general CLIP concepts. However, concepts defined a-priori may not have sufficient predictive power for the task or even be learnable from the available data. As a result, these CBMs often significantl

High-performing vision language models still produce incorrect answers, yet their failure modes are often difficult to explain. To make model internals more accessible and enable systematic debugging, we introduce VisualScratchpad, an interactive interface for visual concept analysis during inference. We apply sparse autoencoders to the vision encoder and link the resulting visual concepts to text tokens via text-to-image attention, allowing us to examine which visual concepts are both captured

Contemporary artificial intelligence research has been organized around two dominant ambitions: productivity, which treats AI systems as tools for accelerating work and economic output, and alignment, which focuses on ensuring that increasingly capable systems behave safely and in accordance with human values. This paper articulates and develops a third, emerging ambition: the use of large language models (LLMs) as scientific instruments for studying human behavior, culture, and moral reasoning.

Neural networks often rely on spurious shortcuts for many epochs before discovering structured representations. However, the mechanism governing when this transition occurs and whether its timing can be predicted remains unclear. Prior work shows that gradient descent converges to low norm solutions and that neural networks exhibit simplicity bias, but neither explains the timescale of the transition from shortcut features to structured representations. We introduce the Norm-Hierarchy Transition

High-performance computing systems are complex machines whose behaviour is governed by the correct functioning of its many subsystems. Among these, the workload scheduler has a crucial impact on the timely execution of the jobs continuously submitted to the computing resources. Making high-quality scheduling decisions is contingent on knowing the duration of submitted jobs before their execution--a non-trivial task for users that can be tackled with Machine Learning. In this work, we devise a wo

This paper bridges neuroscience and artificial intelligence to propose a cortically inspired blueprint for modular perceptual AI. While current monolithic models such as GPT-4V achieve impressive performance, they often struggle to explicitly support interpretability, compositional generalization, and adaptive robustness - hallmarks of human cognition. Drawing on neuroscientific models of cortical modularity, predictive processing, and cross-modal integration, we advocate decomposing perception

Autonomous agents powered by large language models introduce a class of execution-layer vulnerabilities -- prompt injection, retrieval poisoning, and uncontrolled tool invocation -- that existing guardrails fail to address systematically. In this work, we propose the Layered Governance Architecture (LGA), a four-layer framework comprising execution sandboxing (L1), intent verification (L2), zero-trust inter-agent authorization (L3), and immutable audit logging (L4). To evaluate LGA, we construct

Enabling question answering over tables and databases in natural language has become a key capability in the democratization of insights from tabular data sources. These systems first require retrieval of data that is relevant to a given natural language query, for which several methods have been introduced. In this work we present and study a table retrieval mechanism devising fine-grained typed query decomposition and global connectivity-awareness (DCTR), to handle the challenges induced by op

Dataset Condensation (DC) distills knowledge from large datasets into smaller ones, accelerating training and reducing storage requirements. However, despite notable progress, prior methods have largely overlooked the potential of quantization for further reducing storage costs. In this paper, we take the first step to explore post-training quantization in dataset condensation, demonstrating its effectiveness in reducing storage size while maintaining representation quality without requiring exp

Convolutional neural networks (CNNs) have achieved exciting performance in joint segmentation of optic disc and optic cup on single-institution datasets. However, their clinical translation is hindered by two major challenges: limited availability of large-scale, high-quality annotations and performance degradation caused by domain shift during deployment across heterogeneous imaging protocols and acquisition platforms. While unsupervised domain adaptation (UDA) provides a way to mitigate these

The prevailing paradigm in Automated Heuristic Design (AHD) typically relies on the assumption that a single, fixed algorithm can effectively navigate the shifting dynamics of a combinatorial search. This static approach often proves inadequate for Perturbative Heuristics, where the optimal algorithm for escaping local optima depends heavily on the specific search phase. To address this limitation, we reformulate heuristic design as a Non-stationary Bi-level Control problem and introduce DyACE (

Although large vision-language models (LVLMs) have demonstrated remarkable capabilities, they are prone to hallucinations in multi-image tasks. We attribute this issue to limitations in existing attention mechanisms and insufficient cross-image modeling. Inspired by this, we propose a structured hallucination mitigation framework involving Cross-Image Attention calibration and Preference Learning (CAPL). CAPL explicitly enhances inter-image interactions at the architectural level while reinforci

In decentralized Multi-Agent Reinforcement Learning (MARL), steganographic collusion -- where agents develop private protocols to evade monitoring -- presents a critical AI safety threat. Existing defenses, limited to behavioral or reward layers, fail to detect coordination in latent communication channels. We introduce the Dynamic Representational Circuit Breaker (DRCB), an architectural defense operating at the optimization substrate. Building on the AI Mother Tongue (AIM) framework, DRCB util

Large language model-based web agents have shown strong potential in automating web interactions through advanced reasoning and instruction following. While retrieval-based memory derived from historical trajectories enables these agents to handle complex, long-horizon tasks, current methods struggle to generalize across unseen websites. We identify that this challenge arises from the flat memory structures that entangle high-level task logic with site-specific action details. This entanglement

Despite the promise of Retrieval-Augmented Generation in grounding Multimodal Large Language Models with external knowledge, the transition to extensive contexts often leads to significant attention dilution and reasoning hallucinations. The surge in information density causes critical evidence to be submerged by voluminous noise, which complicates the discernment of relevant fragments within a dense input. In this paper, we propose \textbf{Hit-RAG}, a multi-stage preference alignment framework

Source-Free Cross-Domain Few-Shot Learning (SF-CDFSL) focuses on fine-tuning with limited training data from target domains (e.g., medical or satellite images), where Vision-Language Models (VLMs) such as CLIP and SigLIP have shown promising results. Current works in traditional visual models suggest that improving visual discriminability enhances performance. However, in VLM-based SF-CDFSL tasks, we find that \textbf{strengthening visual-modal discriminability actually suppresses VLMs' performa

Safety alignment is critical for deploying large language models (LLMs) in real-world applications, yet most existing approaches rely on large human-annotated datasets and static red-teaming benchmarks that are costly, difficult to scale, and slow to adapt to evolving model behaviors. Moreover, overly conservative safety mechanisms can reduce model usefulness by rejecting sensitive but legitimate queries. We introduce Self-MOA (Self Multi-Objective Alignment), a fully automated framework for ali

Many regulatory and analytic problems require that a prohibited variable influence a decision only through a designated allowable channel -- a conditional-independence requirement that arises in path-specific fairness, the handling of classified information, and the regulation of trading on non-public information, among other settings. Such requirements may be enforced either stratum-by-stratum or, more commonly (and more efficiently), through a single averaged constraint on the conditional effe

Controllable diffusion generation often relies on various heuristics that are seemingly disconnected without a unified understanding. We bridge this gap with Diffusion Controller (DiffCon), a unified control-theoretic view that casts reverse diffusion sampling as state-only stochastic control within (generalized) linearly-solvable Markov Decision Processes (LS-MDPs). Under this framework, control acts by reweighting the pretrained reverse-time transition kernels, balancing terminal objectives ag

Quadruped locomotion provides a natural setting for understanding when model-free learning can outperform model-based control design, by exploiting data patterns to bypass the difficulty of optimizing over discrete contacts and the combinatorial explosion of mode changes. We give a principled analysis of why imitation learning with quadrupeds can be inherently effective in a small data regime, based on the structure of its limit cycles, Poincaré return maps, and local numerical properties of neu

We study post-training linear autoregressive models with outcome and process rewards. Given a context $\boldsymbol{x}$, the model must predict the response $\boldsymbol{y} \in Y^N$, a sequence of length $N$ that satisfies a $γ$ margin condition, an extension of the standard separability to sequences. We prove that on test samples where the base model achieves a non-trivial likelihood $α$, a variant of policy gradient (PG) can achieve likelihood $1 - \varepsilon$ with an essentially minimax optim