The rapid advancement of language models such as This paper presents a robust ensemble approach GPT (Radford et al., 2019) and BERT (Devlin et al., for detecting AI-generated content, with strong 2019) has increased machine-generated content, performance across both English and multilingual raising significant concerns about misinformation tasks. However, significant opportunities remain and academic integrity. Identifying AI-generated for improving model generalization and addressing text becomes more challenging in multilingual contexts, data imbalance, which will be crucial for future where linguistic diversity adds further complexity advancements in this field. The following sections to model generalization. While existing will discuss the dataset, methodology, results, a approaches perform well in English, their effectiveness detailed analysis of the findings, and conclusions decreases when applied to languages with drawn from this study.

This paper presents our approach for Task 3 of the GenAI content detection workshop at COLING-2025, focusing on Cross-Domain Machine-Generated Text (MGT) Detection. We propose an ensemble of fine-tuned transformer models, enhanced by inverse perplexity weighting, to improve classification accuracy across diverse text domains. For Subtask A (Non-Adversarial MGT Detection), we combined a fine-tuned RoBERTa-base model with an OpenAI detector-integrated RoBERTa-base model, achieving an aggregate TPR score of 0.826, ranking 10th out of 23 detectors. In Subtask B (Adversarial MGT Detection), our fine-tuned RoBERTa-base model achieved a TPR score of 0.801, securing 8th out of 22 detectors. Our results demonstrate the effectiveness of inverse perplexity-based weighting for enhancing generalization and performance in both non-adversarial and adversarial MGT detection, highlighting the potential for transformer models in cross-domain AI-generated content detection.

Retrieval Augmented Generation (RAG) systems have been shown to improve the accuracy of Large Language Model (LLM) outputs. However, these models can often achieve low accuracy when applied to new data domains. We introduce the Automatic Local Fine Tuning of Retrieval Augmented Generation models (ALoFTRAG) framework, designed to improve the accuracy of RAG systems on a given domain by training LLMs without manually labeled data or using larger teacher models. By generating and filtering synthetic training data and performing LoRA fine-tuning, ALoFTRAG improves citation and answer accuracy across 20 datasets in 26 languages by, on average, 8.3% and 3.0% respectively. Our results demonstrate that ALoFTRAG offers a practical, cost-effective, and data-secure solution for improving RAG accuracy, making it particularly applicable to sensitive domains such as healthcare and finance.

Despite achieving remarkable performance, machine translation (MT) research remains underexplored in terms of translating cultural elements in languages, such as idioms, proverbs, and colloquial expressions. This paper investigates the capability of state-of-the-art neural machine translation (NMT) and large language models (LLMs) in translating proverbs, which are deeply rooted in cultural contexts. We construct a translation dataset of standalone proverbs and proverbs in conversation for four language pairs. Our experiments show that the studied models can achieve good translation between languages with similar cultural backgrounds, and LLMs generally outperform NMT models in proverb translation. Furthermore, we find that current automatic evaluation metrics such as BLEU, CHRF++ and COMET are inadequate for reliably assessing the quality of proverb translation, highlighting the need for more culturally aware evaluation metrics.

With the rapid growth of online information, the spread of fake news has become a serious social challenge. In this study, we propose a novel detection framework based on Large Language Models (LLMs) to identify and classify fake news by integrating textual statistical features and deep semantic features. Our approach utilizes the contextual understanding capability of the large language model for text analysis and introduces a hybrid attention mechanism to focus on feature combinations that are particularly important for fake news identification. Extensive experiments on the WELFake news dataset show that our model significantly outperforms existing methods, with a 1.5\% improvement in F1 score. In addition, we assess the interpretability of the model through attention heat maps and SHAP values, providing actionable insights for content review strategies. Our framework provides a scalable and efficient solution to deal with the spread of fake news and helps build a more reliable online information ecosystem.

Large Language Models (LLMs) have revolutionized various applications by generating outputs based on given prompts. However, achieving the desired output requires iterative prompt refinement. This paper presents a novel approach that draws parallels between the iterative prompt optimization process in LLMs and feedback control systems. We iteratively refine the prompt by treating the deviation between the LLM output and the desired result as an error term until the output criteria are met. This process is akin to a feedback control system, where the LLM, despite being non-linear and non-deterministic, is managed using principles from linear feedback control systems. We explore the application of different types of controllers within this framework, providing a mathematical foundation for integrating linear feedback control mechanisms with LLMs.

Alongside the former presidents, family members and US officials you would expect to see at Donald Trump's inauguration, there have also been a host of faces familiar for less traditional reasons. We've seen OpenAI CEO Sam Altman taking selfies with influencer brothers Logan and Jake Paul, and controversial Irish mixed martial arts fighter Conor McGregor chatting to British politician Nigel Farage. Also in attendance are tech billionaires like Meta's Mark Zuckerberg and Amazon's Jeff Bezos, media tycoon Rupert Murdoch and FIFA president Gianni Infantino. We will continue spotting the notable and unusual names among the crowd as the day progresses.

This is a very solid paper all around. The idea to cluster activations and force the network to "stay close" to a restricted set of representations is intuitive, acting as a parsimony constraint that additionally enables interpretability. While solid theoretical motivation and analysis isn't provided, the proposed algorithm feels natural and the experiments are fairly comprehensive and compelling. A broad range of tasks are considered, including unsupervised, fine-grained, and zero-shot learning in addition to standard classification, and the visualizations of cluster structure demonstrate that the cluster centers are meaningful---they convince me that the approach is performing as one would hope. Some specific points: It would be nice to see a table or chart demonstrating the effect of different choices of cluster size on generalization error (it is mentioned that cross-validation was used, but it would still be good to get a feel for the sensitivity with respect to choice of cluster size).

Figure 1: Training models to optimize test-time compute and learn "how to discover" correct responses, as opposed to the traditional learning paradigm of learning "what answer" to output. The major strategy to improve large language models (LLMs) thus far has been to use more and more high-quality data for supervised fine-tuning (SFT) or reinforcement learning (RL). Unfortunately, it seems this form of scaling will soon hit a wall, with the scaling laws for pre-training plateauing, and with reports that high-quality text data for training maybe exhausted by 2028, particularly for more difficult tasks, like solving reasoning problems which seems to require scaling current data by about 100x to see any significant improvement. The current performance of LLMs on problems from these hard tasks remains underwhelming (see example). There is thus a pressing need for data-efficient methods for training LLMs that extend beyond data scaling and can address more complex challenges.

Copilot, a tool built on top of OpenAI's large language models and launched by Microsoft-backed GitHub in 2022, is now used by millions of developers around the world. "Today, more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers," Alphabet CEO Sundar Pichai claimed on an earnings call in October: "This helps our engineers do more and move faster." Expect other tech companies to catch up, if they haven't already. A bunch of new startups have entered this buzzy market too. Newcomers such as Zencoder, Merly, Cosine, Tessl (valued at 750 million within months of being set up), and Poolside (valued at 3 billion before it even released a product) are all jostling for their slice of the pie.