position-paper

A survey and position paper arguing that embedding research across modalities is converging via contrastive learning and Transformers, proposing a framework of coverage, comparability, and compositionality with transformation, alignment, and joint embedding as paths toward omnimodal unification.

一篇 Hugging Face 与 Data & Society 合作的立场论文，指出当前 AI agent 设计中的人工监督（HITL）不仅因工作量与复杂度难以有效执行，还会导致监督者技能退化，呼吁在设计与组织层面同等重视监督者的认知需求。

This position paper outlines five major shortcomings of existing faithfulness metrics for explainable AI feature importance and proposes four research avenues to improve their reliability, arguing that XAI explanations currently lack rigorous anchors for quantifying fidelity to the underlying model.

A position paper arguing that AI development should move beyond treating a single ground truth as definitive, instead modeling the space of plausible human judgments while distinguishing meaningful ambiguity from annotation noise.

This position paper argues that offline RL should shift from learning static deployment policies to learning adaptive policy priors that can improve through subsequent online interaction, proposing a framework called Adaptive Offline RL (AORL) and highlighting Bayesian offline RL as a principled approach to preserving epistemic uncertainty.

This position paper analyzes the reproducibility crisis in machine learning—quantifying unavailable code and unreproducible results—and proposes concrete measures to strengthen result checkability and verifiability in ML research.

Google DeepMind's paper proposes levels to operationalize progress towards AGI, prompting discussion on current rankings.

This paper examines language models as linguistic artifacts, using Italian models to explore their training and questioning their representation of language. It argues for a distinction between technical products and research tools in NLP.

This position paper argues that improving machine learning peer review requires enforceable procedural safeguards and a spendable credit system, such as OpenReview Points, to incentivize good reviewing and limit submission volume.

This position paper argues that AI safety research must address AI Lock-In, the phenomenon where excessive reliance on AI leads to human deskilling and systemic vulnerabilities, and provides guidance on mitigation strategies.

This position paper argues that AI evaluation should pivot to assessing human-AI teams rather than superhuman performance to foster better societal outcomes.

This position paper argues that AI systems used in high-stakes decision-making should reason similarly to their users and faithfully communicate that reasoning, and outlines a research agenda for achieving such 'cognitively-aligned AI'.

This position paper argues that modern AI alignment techniques, though designed to prevent harmful outputs, are dual-use technologies that can be misused for censorship and manipulation, and urges the community to address this risk.

This position paper argues that AI reasoning lacks clear operational definitions, undermining evaluation validity, and proposes defining reasoning as a learnable rule-based process with a checklist for research best practices.

This position paper argues that many LLM failures stem from evaluating outputs independently and proposes a self-consistency framework that treats diverse techniques as special cases of consistency optimization.

A position paper arguing that large language models have fundamental limitations, using the metaphor 'can't jump' to highlight gaps in reasoning or generalization.

This position paper argues that long-horizon benchmark failures must be compared against baselines built from matched short stages, introducing the 'horizon residual' metric to distinguish task size from task difficulty in LLM agent evaluation.

This position paper argues that language model evaluation scores should be treated as perishable knowledge claims, not ground truth, and proposes explicit metadata such as formality tier, scope declaration, and expiration date to counter 'trust inflation' caused by averaging weak and strong signals.

This position paper argues against the claim that natural language can fully replace formal languages such as programming languages, proposing an information-theoretic specificity framework and proving a crossover theorem showing formal languages are better for high-specificity tasks.

This position paper argues for shifting from reactive AI flywheel maintenance (patching errors as they occur) to a proactive test-driven approach that maps feedback to a test space of task conditions, showing mathematically that the proactive method achieves better long-term scaling with fewer iterations.