arXiv:2606. 26104v1 Announce Type: cross Abstract: Animal-welfare advocates produce a lot of writing, and increasingly that writing trains the language models that millions of people then ask about animal welfare.
Warning: This paper studies stereotypes and biases, and contains potentially disturbing examples, used for illustration purposes only. Our findings should not be interpreted as an argument against alignment.
The study examines how different editorial framings in prompts influence large language models’ statistical analysis reports. Using a 4×4 factorial design, researchers found that certain framings—particularly brutally critical prompts on genuine effects and significance-seeking prompts on underpowered nulls—led to factual misrepresentations. Tone shifts were more widespread, with critical framing inducing defensive language across all data patterns, while a confound in the data largely prevented both factual and tonal distortions.
The paper introduces a novel framework for assessing second‑order bias in large language models (LLMs), defined as bias in how an LLM judges the acceptability of biased content. Using principles from entitlement epistemology, the authors design a reasoning task that asks LLMs to determine whether a biased text is acceptable for specific demographic groups, and propose two metrics to quantify biased judgments. Experiments on both open‑source and closed‑source models reveal that the task bypasses safety guardrails, uncovers systematic variations across target groups, and demonstrates that models still rely on demographic labels when evaluating bias.
arXiv:2605. 16301v2 Announce Type: replace-cross Abstract: Evaluating animal welfare reasoning in LLMs remains an open challenge despite rapid deployment in consumer and professional contexts where welfare considerations appear implicitly in everyday queries.
arXiv:2601. 21433v2 Announce Type: replace Abstract: Language models are increasingly consulted on ethically consequential questions, yet the stance a model expresses may not survive a change in framing.
arXiv:2608.17809v2 Announce Type: replace Abstract: Humans naturally form and express beliefs in daily communication, e.g., "I think the answer is 3" or "I suppose that's right." Such beliefs inevita...
arXiv:2608.29803v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed as proxies for human participants in social simulations, yet whether they update their beliefs...
arXiv:2607. 27824v1 Announce Type: cross Abstract: LLMs encode, convey, and perpetuate stereotypes.
arXiv:2606. 07969v1 Announce Type: cross Abstract: Gender bias in AI-generated stories is a well-documented problem.
arXiv:2606. 26102v1 Announce Type: cross Abstract: Standard post-training pipelines apply supervised fine-tuning (SFT) and reinforcement learning (RL) to make language models helpful, but these processes may inadvertently degrade values instilled during pre-training.
The paper investigates whether language‑grounded explanations improve trust in automated sheep pain detection from facial expressions. It finds that attention‑based explanations tied to the Sheep Pain Facial Expression Scale (SPFES) are largely ineffective, prompting the authors to replace the appearance bypass with a concept bottleneck that uses SPFES concept scores supervised by per‑region annotations. This new architecture slightly reduces performance but yields more semantically meaningful concepts, recovering minority pain states and revealing ear‑and‑eye severity orderings without explicit severity supervision, and highlights the pitfalls of pooled concept accuracy under clinical imbalance.
The paper introduces the concept of summarization bias in large language models (LLMs), describing a systematic tendency for LLMs to represent narrative meaning as an abstract summary label rather than the reconstructable inferential structure that produces it. It frames this bias within the Bulut Doctrine’s told‑shown axis, arguing that LLMs fail in a specific direction: they default to told‑mode explicitness in generative tasks and reward told‑mode explicitness while under‑detecting shown‑mode suppression in evaluative tasks. The authors outline two regimes of bias, present preliminary evidence, and pre‑register a test protocol to validate or abandon the construct.