As artificial intelligence systems increasingly inform high-stakes decisions across sectors, transparency has become foundational to responsible and trustworthy AI implementation. Leveraging our role as a leading institute in advancing AI research and enabling industry adoption, we present key insights and lessons learned from practical interpretability applications across diverse domains. This paper offers actionable strategies and implementation guidance tailored to organizations at varying stages of AI maturity, emphasizing the integration of interpretability as a core design principle rather than a retrospective add-on.

Large language models (LLMs) increasingly serve as educational tools, yet evaluating their teaching capabilities remains challenging due to the resource-intensive, context-dependent, and methodologically complex nature of teacher-student interactions. We introduce EducationQ, a multi-agent dialogue framework that efficiently assesses teaching capabilities through simulated dynamic educational scenarios, featuring specialized agents for teaching, learning, and evaluation. Testing 14 LLMs across major AI Organizations (OpenAI, Meta, Google, Anthropic, and others) on 1,498 questions spanning 13 disciplines and 10 difficulty levels reveals that teaching effectiveness does not correlate linearly with model scale or general reasoning capabilities - with some smaller open-source models outperforming larger commercial counterparts in teaching contexts. This finding highlights a critical gap in current evaluations that prioritize knowledge recall over interactive pedagogy. Our mixed-methods evaluation, combining quantitative metrics with qualitative analysis and expert case studies, identifies distinct pedagogical strengths employed by top-performing models (e.g., sophisticated questioning strategies, adaptive feedback mechanisms). Human expert evaluations show 78% agreement with our automated qualitative analysis of effective teaching behaviors, validating our methodology. EducationQ demonstrates that LLMs-as-teachers require specialized optimization beyond simple scaling, suggesting next-generation educational AI prioritize targeted enhancement of specific pedagogical effectiveness.

Autonomous Artificial Intelligence (AI) has many benefits. It also has many risks. In this work, we identify the 3 levels of autonomous AI. We are of the position that AI must not be fully autonomous because of the many risks, especially as artificial superintelligence (ASI) is speculated to be just decades away. Fully autonomous AI, which can develop its own objectives, is at level 3 and without responsible human oversight. However, responsible human oversight is crucial for mitigating the risks. To ague for our position, we discuss theories of autonomy, AI and agents. Then, we offer 12 distinct arguments and 6 counterarguments with rebuttals to the counterarguments. We also present 15 pieces of recent evidence of AI misaligned values and other risks in the appendix.

Due to growing privacy concerns, machine unlearning, which aims at enabling machine learning models to ``forget" specific training data, has received increasing attention. Among existing methods, influence-based unlearning has emerged as a prominent approach due to its ability to estimate the impact of individual training samples on model parameters without retraining. However, this approach suffers from prohibitive computational overhead arising from the necessity to compute the Hessian matrix and its inverse across all training samples and parameters, rendering it impractical for large-scale models and scenarios involving frequent data deletion requests. This highlights the difficulty of forgetting. Inspired by cognitive science, which suggests that memorizing is easier than forgetting, this paper establishes a theoretical link between memorizing (incremental learning) and forgetting (unlearning). This connection allows machine unlearning to be addressed from the perspective of incremental learning. Unlike the time-consuming Hessian computations in unlearning (forgetting), incremental learning (memorizing) typically relies on more efficient gradient optimization, which supports the aforementioned cognitive theory. Based on this connection, we introduce the Influence Approximation Unlearning (IAU) algorithm for efficient machine unlearning from the incremental perspective. Extensive empirical evaluations demonstrate that IAU achieves a superior balance among removal guarantee, unlearning efficiency, and comparable model utility, while outperforming state-of-the-art methods across diverse datasets and model architectures. Our code is available at https://github.com/Lolo1222/IAU.

Traditional topic models are effective at uncovering latent themes in large text collections. However, due to their reliance on bag-of-words representations, they struggle to capture semantically abstract features. While some neural variants use richer representations, they are similarly constrained by expressing topics as word lists, which limits their ability to articulate complex topics. We introduce Mechanistic Topic Models (MTMs), a class of topic models that operate on interpretable features learned by sparse autoencoders (SAEs). By defining topics over this semantically rich space, MTMs can reveal deeper conceptual themes with expressive feature descriptions. Moreover, uniquely among topic models, MTMs enable controllable text generation using topic-based steering vectors. To properly evaluate MTM topics against word-list-based approaches, we propose \textit{topic judge}, an LLM-based pairwise comparison evaluation framework. Across five datasets, MTMs match or exceed traditional and neural baselines on coherence metrics, are consistently preferred by topic judge, and enable effective steering of LLM outputs.

This article redefines arbitrariness not as a normative flaw or a symptom of domination, but as a foundational functional mechanism structuring human systems and interactions. Diverging from critical traditions that conflate arbitrariness with injustice, it posits arbitrariness as a semiotic trait: a property enabling systems - linguistic, legal, or social - to operate effectively while withholding their internal rationale. Building on Ferdinand de Saussure's concept of l'arbitraire du signe, the analysis extends this principle beyond language to demonstrate its cross-domain applicability, particularly in law and social dynamics. The paper introduces the "Motivation -> Constatability -> Contestability" chain, arguing that motivation functions as a crucial interface rendering an act's logic vulnerable to intersubjective contestation. When this chain is broken through mechanisms like "immotivization" or "Conflict Lateralization" (exemplified by "the blur of the wolf drowned in the fish"), acts produce binding effects without exposing their rationale, thus precluding justiciability. This structural opacity, while appearing illogical, is a deliberate design protecting authority from accountability. Drawing on Shannon's entropy model, the paper formalizes arbitrariness as A = H(L|M) (conditional entropy). It thereby proposes a modern theory of arbitrariness as a neutral operator central to control as well as care, an overlooked dimension of interpersonal relations. While primarily developed through human social systems, this framework also illuminates a new pathway for analyzing explainability in advanced artificial intelligence systems.

Agentic search, as a more autonomous and adaptive paradigm of retrieval augmentation, is driving the evolution of intelligent search systems. However, existing evaluation frameworks fail to align well with the goals of agentic search. First, the complex queries commonly used in current benchmarks often deviate from realistic user search scenarios. Second, prior approaches tend to introduce noise when extracting ground truth for end-to-end evaluations, leading to distorted assessments at a fine-grained level. Third, most current frameworks focus solely on the quality of final answers, neglecting the evaluation of the iterative process inherent to agentic search. To address these limitations, we propose RAVine -- a Reality-Aligned eValuation framework for agentic LLMs with search. RAVine targets multi-point queries and long-form answers that better reflect user intents, and introduces an attributable ground truth construction strategy to enhance the accuracy of fine-grained evaluation. Moreover, RAVine examines model's interaction with search tools throughout the iterative process, and accounts for factors of efficiency. We benchmark a series of models using RAVine and derive several insights, which we hope will contribute to advancing the development of agentic search systems. The code and datasets are available at https://github.com/SwordFaith/RAVine.

Large Language Models (LLMs) increasingly shape public discourse, yet most evaluations of political and economic bias have focused on high-resource, Western languages and contexts. This leaves critical blind spots in low-resource, multilingual regions such as Pakistan, where linguistic identity is closely tied to political, religious, and regional ideologies. We present a systematic evaluation of political bias in 13 state-of-the-art LLMs across five Pakistani languages: Urdu, Punjabi, Sindhi, Pashto, and Balochi. Our framework integrates a culturally adapted Political Compass Test (PCT) with multi-level framing analysis, capturing both ideological stance (economic/social axes) and stylistic framing (content, tone, emphasis). Prompts are aligned with 11 socio-political themes specific to the Pakistani context. Results show that while LLMs predominantly reflect liberal-left orientations consistent with Western training data, they exhibit more authoritarian framing in regional languages, highlighting language-conditioned ideological modulation. We also identify consistent model-specific bias patterns across languages. These findings show the need for culturally grounded, multilingual bias auditing frameworks in global NLP.

Fairness in machine learning research is commonly framed in the context of classification tasks, leaving critical gaps in regression. In this paper, we propose a novel approach to measure intersectional fairness in regression tasks, going beyond the focus on single protected attributes from existing work to consider combinations of all protected attributes. Furthermore, we contend that it is insufficient to measure the average error of groups without regard for imbalanced domain preferences. Accordingly, we propose Intersectional Divergence (ID) as the first fairness measure for regression tasks that 1) describes fair model behavior across multiple protected attributes and 2) differentiates the impact of predictions in target ranges most relevant to users. We extend our proposal demonstrating how ID can be adapted into a loss function, IDLoss, that satisfies convergence guarantees and has piecewise smooth properties that enable practical optimization. Through an extensive experimental evaluation, we demonstrate how ID allows unique insights into model behavior and fairness, and how incorporating IDLoss into optimization can considerably improve single-attribute and intersectional model fairness while maintaining a competitive balance in predictive performance.

Large Language Models (LLMs) have achieved remarkable results on a range of standardized tests originally designed to assess human cognitive and psychological traits, such as intelligence and personality. While these results are often interpreted as strong evidence of human-like characteristics in LLMs, this paper argues that such interpretations constitute an ontological error. Human psychological and educational tests are theory-driven measurement instruments, calibrated to a specific human population. Applying these tests to non-human subjects without empirical validation, risks mischaracterizing what is being measured. Furthermore, a growing trend frames AI performance on benchmarks as measurements of traits such as ``intelligence'', despite known issues with validity, data contamination, cultural bias and sensitivity to superficial prompt changes. We argue that interpreting benchmark performance as measurements of human-like traits, lacks sufficient theoretical and empirical justification. This leads to our position: Stop Evaluating AI with Human Tests, Develop Principled, AI-specific Tests instead. We call for the development of principled, AI-specific evaluation frameworks tailored to AI systems. Such frameworks might build on existing frameworks for constructing and validating psychometrics tests, or could be created entirely from scratch to fit the unique context of AI.