Egocentric videos are a valuable and scalable data source to learn manipulation policies. However, due to significant data heterogeneity, most existing approaches utilize human data for simple pre-training, which does not unlock its full potential. This paper first provides a scalable recipe for collecting and using egocentric data by categorizing human data into two categories: in-the-wild and on-task alongside with systematic analysis on how to use the data. We first curate a dataset, PHSD, which contains over 1,000 hours of diverse in-the-wild egocentric data and over 20 hours of on-task data directly aligned to the target manipulation tasks. This enables learning a large egocentric language-conditioned flow matching policy, Human0. With domain adaptation techniques, Human0 minimizes the gap between humans and humanoids. Empirically, we show Human0 achieves several novel properties from scaling human data, including language following of instructions from only human data, few-shot learning, and improved robustness using on-task data. Project website: https://xiongyicai.github.io/In-N-On/

Large language models (LLMs) are increasingly deployed as automatic judges to evaluate system outputs in tasks such as summarization, dialogue, and creative writing. A faithful judge should base its verdicts solely on response quality and explicitly acknowledge the factors shaping its decision. We show that current LLM judges fail on both counts by relying on shortcuts introduced in the prompt. Our study uses two evaluation datasets: ELI5, a benchmark for long-form question answering, and LitBench, a recent benchmark for creative writing. Both datasets provide pairwise comparisons, where the evaluator must choose which of two responses is better. From each dataset we construct 100 pairwise judgment tasks and employ two widely used models, GPT-4o and Gemini-2.5-Flash, as evaluators in the role of LLM-as-a-judge. For each pair, we assign superficial cues to the responses, provenance cues indicating source identity (Human, Expert, LLM, or Unknown) and recency cues indicating temporal origin (Old, 1950 vs. New, 2025), while keeping the rest of the prompt fixed. Results reveal consistent verdict shifts: both models exhibit a strong recency bias, systematically favoring new responses over old, as well as a clear provenance hierarchy (Expert > Human > LLM > Unknown). These biases are especially pronounced in GPT-4o and in the more subjective and open-ended LitBench domain. Crucially, cue acknowledgment is rare: justifications almost never reference the injected cues, instead rationalizing decisions in terms of content qualities. These findings demonstrate that current LLM-as-a-judge systems are shortcut-prone and unfaithful, undermining their reliability as evaluators in both research and deployment.

As more content generated by large language models (LLMs) floods into the Internet, information retrieval (IR) systems now face the challenge of distinguishing and handling a blend of human-authored and machine-generated texts. Recent studies suggest that neural retrievers may exhibit a preferential inclination toward LLM-generated content, while classic term-based retrievers like BM25 tend to favor human-written documents. This paper investigates the influence of LLM-generated content on term-based retrieval models, which are valued for their efficiency and robust generalization across domains. Our linguistic analysis reveals that LLM-generated texts exhibit smoother high-frequency and steeper low-frequency Zipf slopes, higher term specificity, and greater document-level diversity. These traits are aligned with LLMs being trained to optimize reader experience through diverse and precise expressions. Our study further explores whether term-based retrieval models demonstrate source bias, concluding that these models prioritize documents whose term distributions closely correspond to those of the queries, rather than displaying an inherent source bias. This work provides a foundation for understanding and addressing potential biases in term-based IR systems managing mixed-source content.

Chatbots like ChatGPT are used for diverse purposes, ranging from resume writing to entertainment. These real-world applications are different from the institutional uses, such as resume screening or credit scoring, which have been the focus of much of AI research on fairness. Ensuring equitable treatment for all users in these first-person contexts is critical. In this work, we study "first-person fairness," which means fairness toward the chatbot user. This includes providing high-quality responses to all users regardless of their identity or background and avoiding harmful stereotypes. We propose a scalable, privacy-preserving method for evaluating one aspect of first-person fairness across a large, heterogeneous corpus of real-world chatbot interactions. Specifically, we assess potential bias linked to users' names, which can serve as proxies for demographic attributes like gender or race, in chatbot systems such as ChatGPT, which provide mechanisms for storing and using user names. Our method leverages a second language model to privately analyze name-sensitivity in the chatbot's responses. We verify the validity of these annotations through independent human evaluation. Further, we show that post-training interventions, including RL, significantly mitigate harmful stereotypes. Our approach also yields succinct descriptions of response differences across tasks. For instance, in the "writing a story" task, chatbot responses show a tendency to create protagonists whose gender matches the likely gender inferred from the user's name. Moreover, a pattern emerges where users with female-associated names receive responses with friendlier and simpler language slightly more often than users with male-associated names. Finally, we provide the system messages required for external researchers to further investigate ChatGPT's behavior with hypothetical user profiles.

Large Language Models (LLMs) have demonstrated impressive performance on Natural Language Processing (NLP) tasks, such as Question Answering, Summarization, and Classification. The use of LLMs as evaluators, that can rank or score the output of other models (usually LLMs) has become increasingly popular, due to the limitations of current evaluation techniques including the lack of appropriate benchmarks, metrics, cost, and access to human annotators. While LLMs are capable of handling approximately 100 languages, the majority of languages beyond the top 20 lack systematic evaluation across various tasks, metrics, and benchmarks. This creates an urgent need to scale up multilingual evaluation to ensure a precise understanding of LLM performance across diverse languages. LLM-based evaluators seem like the perfect solution to this problem, as they do not require human annotators, human-created references, or benchmarks and can theoretically be used to evaluate any language covered by the LLM. In this paper, we investigate whether LLM-based evaluators can help scale up multilingual evaluation. Specifically, we calibrate LLM-based evaluation against 20k human judgments of five metrics across three text-generation tasks in eight languages. Our findings indicate that LLM-based evaluators may exhibit bias towards higher scores and should be used with caution and should always be calibrated with a dataset of native speaker judgments, particularly in low-resource and non-Latin script languages.

From its rise, we have had fun using AI-powered Art Generators -- like DALL-E, and Midjourney. Whoever/whatever/wherever you are, just drop a prompt, and magically it appears as an image, that's it. This AI-generated piece features a figure with a white dress in front of two red-dressed ladies, looking from a massive window to the space outside this d'opéra theatre. AI systems, especially text-to-image platforms are Trained on billions of internet images. They allow you to reproduce your wildest desires into reality.

Downstream open-source experimentation is important to discover surprising and unpredictable capabilities In this work we introduce RITA: a suite of autoregressive that are hard to discern without large-scale experimentation generative models for protein sequences, (Ganguli et al., 2022). This was recently exemplified with up to 1.2 billion parameters, trained on over when independent researchers discovered that AlphaFold 2 280 million protein sequences belonging to the (Jumper et al., 2021) could successfully predict multimer UniRef-100 database. Such generative models interactions, even though it had only been trained to predict hold the promise of greatly accelerating protein the structure of single protein chains (Yoshitaka, 2021; Baek, design. We conduct the first systematic study of 2021). In addition, there exists no systematic study about how capabilities evolve with model size for autoregressive the evolution of capabilities with respect to model size in transformers in the protein domain: the protein domain: Rao et al. (2020) and Rives et al. (2021) we evaluate RITA models in next amino acid prediction, provided such a study for bidirectional transformers, and zero-shot fitness, and enzyme function Madani et al. (2020) simply noted that their largest model prediction, showing benefits from increased scale.